CN1484756A - Coding device and decoding device - Google Patents

Abstract

A sound data input part (310) of the encoding device (300) cuts 4096 pieces of successive sound data out of a sound data sequence and a conversion part (320) converts the cut sound data into spectrum data on the frequency axis. A data separation part (330) separates the spectrum data into a low-frequency part and a high-frequency part based upon 11.025 kHz as a border. The low-frequency part spectrum data are quantized and encoded by a 1st quantization part (340) and an encoding part (350) as usual. A 2nd encoding part (345) generates auxiliary information showing a feature of a high-frequency part frequency spectrum and a 2nd encoding part (355) encodes the auxiliary information. A stream output part (390) puts together and output the code obtained by the 1st encoding part (350) and the code obtained by the 2nd encoding part (355). Here, f1 is less than a half of the sampling frequency f2 when the sound data sequence was generated.

Description

Encoding device, decoding device and voice data distribution system

Technical field

Thereby the present invention relates to the regenerate technology of high-quality sound of a kind of compressed/encoded and expansion/decoded audio signal.

Background technology

In recent years, the compressed/encoded of multiple sound signal and the method for expansion/decoding have appearred.MPEG-2 Advanced Audio Coding (being called as " MPEG-2 AAC " or " AAC " later on) is exactly a kind of such technology.(referring to " IS13818-7 " (MPEG-2 Advanced Audio Coding, AAC) author M.Bosi etc., in April, 1997).

Accompanying drawing 1 is a block scheme, shows according to the encoding device of traditional AAC method and the functional structure of decoding device.

Encoding device 1000 is the equipment according to AAC coding method compression and coding input audio signal, this equipment comprises an A/D converter 1050, a sound signal input block 1100, a converter unit 1200,1400, one coding units of a quantifying unit and a stream output unit 1900.

A/D converter 1050 is sampled to input signal with the sample frequency of for example 22.05kHz, and converts simulated audio signal to a digital audio-frequency data string.In the time of 1024 samplings of voice data string of having read input signal whenever audio frequency input block 1100 (these 1024 samplings are called as one " frame " later on), it is separated into 2048 data samplings with the voice data string, these sampled datas have two groups of respectively samplings of half, are used for the frame (512) that is obtained before and after this frame is capped.

1200 pairs of 2048 data samplings that separated by voice data input block 1100 in time domain of converter unit are carried out and are improved discrete cosine transform (MDCT), and it is transformed into the frequency spectrum data of frequency field.1024 samplings of frequency spectrum data, i.e. half of the frequency spectrum data that conversion obtains, expression 11.025kHz or littler regeneration bandwidth, and be divided into a plurality of groups.Thereby each group all is set and comprises one or more frequency spectrum data sampling.And, the critical bandwidth of each group simulating human sense of hearing, it is called as one " scale factor (scale factorband) ".

Quantifying unit 1400 frequency spectrum data that quantization transform unit 1200 is produced in scale factor utilizes a normalization factor that it is quantified as predetermined number of bits in each scale factor.This normalization factor is called as one " scale factor ".And the result who utilizes each scale factor that each frequency spectrum data is quantized is called as one " quantized value ".Coding unit 1500 is exactly each scale factor according to the data that huffman coding coded quantization unit 1400 quantizes, and the frequency spectrum data that uses this scale factor to quantize.

The coded signal that stream output unit 1900 will result from coding unit 1500 is transformed into an AAC bitstream format, and with its output.Be sent to encoding device 2000 from the bit stream of encoding device 1000 outputs by transmission medium or recording medium.

Decoding device 2000 is the equipment of decoding and coding equipment 1000 bitstream encoded, this equipment comprises a stream input block 2100, a decoding unit 2200, a de-quantization unit 2300,2800, one voice data output units 2900 of an inverse transformation block and a D/A converter 2950.

Stream input block 2100 is by transmission medium or pass through recording medium received code equipment 1000 bitstream encoded, and reads coded signal from the bit stream that receives.The signal of huffman coding produces quantized data thereby decoding unit 200 is decoded then.

The quantized data of de-quantization unit 2300 proportion of utilization factor pair decoding units, 2200 decodings carries out de-quantization.The contrary discrete cosine transform (IMDCT) that improves is carried out in 1024 samplings of the frequency spectrum data of the frequency field that 2800 pairs of de-quantization unit of inverse transformation block 2300 produce, and it is transformed into the voice data of 1024 samplings of time domain.Voice data output unit 2900 order merges the voice data of 1024 samplings of the time domain that inverse transformation block 2800 produces, and exports the voice data of 1024 samplings of this group one by one with time sequencing.D/A converter 2950 converts digital audio-frequency data to analog audio data with the sample frequency of 22.05kHz.

In above-mentioned encoding device 1000 and decoding device 2000 according to traditional AAC standard, each sampled data can be compressed to 1 bit or littler.In addition, because 1024 frequency spectrum data samplings in low-frequency band are encoded, these samplings have higher sense of hearing priority, so voice data can be reproduced with relative higher quality, low-frequency band is expressed as 11.025kHz or littler regeneration bandwidth, and it is sample frequency half.

But, in encoding device 1000 and decoding device 2000, because sample frequency is 22.05kHz, with the data that do not have in the frequency spectrum data that is encoded in the above bandwidth of 11.025kHz according to traditional AAC method (prior art 1).Therefore just exist a problem, expectation hears that the requirement that is included in the higher-quality sound in the above bandwidth of 11.025kHz just can't be satisfied.

In order to address this problem, consider the sample frequency of the D/A converter 2950 that is applied to A/D converter 1050 in the encoding device 1000 and decoding device 2000 among Fig. 1 is brought up to a times of 22.05kHz, be exactly 44.1kHz (prior art 2).

But, if sample frequency is 44.1kHz, being positioned at 512 frequency spectrum data samplings that are higher than the above higher-frequency bandwidth of 11.025kHz can be encoded, keep a compression factor simultaneously, but the frequency spectrum data with sense of hearing higher priority in low-frequency band will be reduced to half, is exactly 512 samplings.In other words, the number of the frequency spectrum data in sample frequency and the low-frequency band is the relation that is conversion (trade-off) alternately, and the two can't be enhanced simultaneously.Therefore be exactly that whole sound quality is by deterioration with regard to having produced another problem.

This problem all can occur in according to the encoding device of other method (for example MP3.AC3 etc.) and decoding device.

The present invention is designed to solve the above problems, and the purpose of this invention is to provide and a kind ofly can realize that high-quality sound reproduction does not increase the encoding device and the decoding device of the data volume behind the coding simultaneously.

Summary of the invention

To achieve these goals, encoding device according to the present invention is a kind of encoding device that can coding audio data, comprising: a separative element is used for audio data stream is separated into the continuant audio data of a fixed number; A converter unit, the voice data that is used for separating is transformed into the frequency spectrum data of frequency field; A division unit is used for being divided into the frequency spectrum data that converter unit obtains at the frequency spectrum data of frequency f 1Hz or lower low-frequency band and being higher than frequency spectrum data in the high frequency bandwidth of f1Hz; A low-frequency band coding unit is used to quantize frequency spectrum data and this quantized data of encoding of division in the low-frequency band; A sub-information generation unit is used to produce the spectral characteristic of the high frequency band that shows the frequency spectrum data that comes from the division in the high frequency band; A high frequency band coding unit is used to the sub-information of encoding and producing; And output unit, be used for the coding of integrated (integrate) low-frequency band coding unit generation and the coding that the high frequency band coding unit produces, and export integrated sign indicating number, wherein f1 be sample frequency f2 half or littler, produce the voice data string at frequency f 2 places.

In encoding device according to the present invention, the frequency spectrum data in f1 or lower low-frequency band in the converter unit output voice data that separative element separated, simultaneously, output is higher than the frequency spectrum data in the high frequency band of f1.Be divided frequency spectrum data in the low-frequency band of dividing elements and be quantized and encode, the frequency spectrum data in the high frequency band is encoded into the sub-information of the characteristic of expression high frequency band.The sub-information that high frequency band coding unit coding produces.Therefore, thus the sound signal high-quality sound of regenerating that can be encoded in the high frequency band, in addition the sound signal in the low frequency bandwidth can be encoded to the same mode of down-sampling (down-sampling), and do not increase the total amount of data basically.

Here, f1 is f2/4, and converter unit can be transformed into 0～2 * f1Hz with voice data, and division unit can be divided into the frequency spectrum data of 0～2 * f1Hz at the frequency spectrum data of frequency f 1Hz or lower low-frequency band and at the frequency spectrum data in the high frequency band of 2 * f1Hz more than the f1Hz.Perhaps, frequency spectrum data in frequency f 1Hz or lower low-frequency band comprises n frequency spectrum data sampling, separative element can be separated into the voice data string and be used to produce 2 * n frequency spectrum data sampling requisite number purpose voice data, converter unit can be transformed into the voice data of division 2 * n frequency spectrum data sampling, and division unit can be divided into 2 * n frequency spectrum data sampling n the frequency spectrum data sampling that n frequency spectrum data taken a sample and high frequency band is interior in the low-frequency band.Perhaps, separative element can be separated into the voice data string 2 * n frequency spectrum data sampling, the sampling of these frequency spectrum datas comprises with the corresponding n of a frame voice data sampling and two groups of n/2 voice datas before this frame and in two frames of vicinity afterwards as a coding unit takes a sample, converter unit is carried out MDCT to 2 * n the voice data sampling that separates, and it is transformed into the frequency spectrum of 0～2 * f1Hz that comprises 2 * n frequency spectrum data sampling.

And, decoding device according to the present invention is one and can decodes through the equipment of the coded data of recording medium or transmission medium input, comprise: an extracting unit is used for extracting the coded data of the low-frequency band that is included in coded data and the coded data of high frequency band; A low-frequency band de-quantization unit is used to decode and the coded data of the low-frequency band that de-quantization is extracted by extracting unit, thereby output is at the frequency spectrum data of frequency f 1Hz or lower low-frequency band; A sub-information decoding unit, the coded data of the high frequency band that extracts by extracting unit of being used to decode, thus produce the sub-information of the frequency spectrum data characteristic that shows high frequency band; A high frequency band de-quantization unit is used for the sub-information based on the generation of sub-information decoding unit, the frequency spectrum data in the output high frequency band; An integrated unit is used for the frequency spectrum data of the high frequency band of the frequency spectrum data of low-frequency band of integrated low-frequency band de-quantization unit output and the output of high frequency band de-quantization unit; An inverse transformation block is used for the voice data that the frequency spectrum data inversion that integrated unit is integrated changes time domain into; A voice data output unit is used for the voice data based on the inverse transformation of time sequencing output inverse transformation block.

In decoding device according to the present invention, extracting unit extracts low-frequency band coded data and high frequency band coded data from the coded data of input, the frequency spectrum data of low-frequency band de-quantization unit output frequency f1Hz or lower low-frequency band.The sub-information of sub-information decoding unit decodes, and the output of high frequency band de-quantization unit is based on the frequency spectrum data of the high frequency band of sub-information.Like this, can utilize with classic method much at one the less data of quantity decode and be far longer than the data of classic method, thereby the sound signal high-quality sound of can decodedly regenerating simultaneously.

It should be noted that, the present invention can realize as a kind of communication system of above-mentioned encoding device and decoding device that comprises certainly, also can be as being included in above-mentioned encoding device, a kind of coding method of performed step in the feature unit of decoding device and communication system, a kind of coding/decoding method and communication means are realized, be used as above-mentioned encoding device as a kind of CPU of order, the coded program and the decoding program of the feature unit of decoding device and communication system or step wherein realize, perhaps realizes as a kind of computer readable recording medium storing program for performing that is writing down these programs on it.

Description of drawings

These and other purpose of the present invention, advantage and characteristics will be from becoming clear subsequently with the accompanying drawing illustrated together, and these accompanying drawings have shown one particular embodiment of the present invention.Wherein:

Accompanying drawing 1 is one and shows the block scheme according to the functional structure of the encoding device of traditional AAC method and decoding device.

Accompanying drawing 2 is one and shows the block scheme according to the functional structure of broadcast system of the present invention.

Accompanying drawing 3A and 3B are the synoptic diagram that shows the state variation of handled sound signal in the encoding device shown in the accompanying drawing 2.

Accompanying drawing 4 is operational flowcharts that show in the definite processing of the performed scale factor of first quantifying unit shown in Figure 2.

Accompanying drawing 5 is another operational flowcharts that show in the definite processing of the performed scale factor of first quantifying unit shown in Figure 2.

Accompanying drawing 6 is spectrum waveform figure of the specific embodiment of the sub-information (scale factor) that shows that second quantifying unit shown in the accompanying drawing 2 is produced.

Accompanying drawing 7 is operational flowcharts of sub-information (scale factor) computing that shows that second quantifying unit shown in Figure 2 is performed.

Accompanying drawing 8A～8C is the figure in bit stream zone that shows the sub-information of stream output unit shown in Figure 2 storage.

Accompanying drawing 9A and 9B show the figure of other embodiment in bit stream zone of the sub-information of stream output unit shown in Figure 2 storage.

Accompanying drawing 10A and 10B show the comparison of processing procedure of the encoding device of encoding device shown in Figure 2 and prior art 1.

Accompanying drawing 11A and 11B show the comparison of processing procedure of the encoding device of encoding device shown in Figure 2 and prior art 2.

Accompanying drawing 12 shows the frequency spectrum data of encoding device of encoding device shown in Figure 2 and prior art 1 and 2 and the comparison of characteristic.

Accompanying drawing 13 is to show that the second de-quantization unit shown in Figure 2 copies to 1024 frequency spectrum datas in the low frequency band process flowchart of high frequency band with the forward direction direction.

Accompanying drawing 14 is to show that the second de-quantization unit shown in Figure 2 copies to 1024 frequency spectrum datas in the low frequency band process flowchart of high frequency band with the direction opposite with frequency axis.

Accompanying drawing 15 is spectrum waveform figure of the embodiment of another sub-information (quantized value) of showing that second quantifying unit shown in Figure 2 produces.

Accompanying drawing 16 is operational flowcharts of another sub-information (quantized value) computing of showing that second quantifying unit shown in Figure 2 is performed.

Accompanying drawing 17 is spectrum waveform figure of the embodiment of another sub-information (positional information) of showing that second quantifying unit shown in Figure 2 is produced.

Accompanying drawing 18 is operational flowcharts of another sub-information (positional information) computing of showing that second quantifying unit shown in Figure 2 is performed.

Accompanying drawing 19 is spectrum waveform figure of the embodiment of another sub-information (symbolic information) of showing that second quantifying unit shown in Figure 2 is produced.

Accompanying drawing 20 is operational flowcharts of another sub-information (symbolic information) computing of showing that second quantifying unit shown in Figure 2 is performed.

Accompanying drawing 21A and 21B are the spectrum waveform figure that shows the embodiment that how to produce another sub-information (Copy Info) that second quantifying unit shown in Figure 2 produced.

Accompanying drawing 22 is operational flowcharts of another sub-information (Copy Info) computing of showing that second quantifying unit shown in Figure 2 is performed.

Accompanying drawing 23 is the spectrum waveform figure that show the embodiment that how to produce another sub-information (Copy Info) that second quantifying unit shown in Figure 2 produced.

Accompanying drawing 24 is operational flowcharts of another sub-information (Copy Info) computing of showing that second quantifying unit shown in Figure 2 is performed.

Embodiment

Describe the present invention below with reference to accompanying drawings and be applied to situation as the embodiment of a kind of broadcast system of voice data distribution system.

Accompanying drawing 2 is the block schemes that show according to the functional structure of broadcast system of the present invention.

According to the broadcast system 1 of present embodiment, as shown in Figure 2, be placed in the broadcasting station, this system comprises the encoding device 300 of a coding input audio signal, and the decoding device 400 of the sound signal bit stream of decoding and coding equipment 300 codings.

(encoding device 300)

Encoding device 300 is when receiving a sound signal, this sound signal of encoding, this encoding device 300 comprises 310, one converter units 320 of 305, one voice data input blocks of an A/D converter, a data division unit 330,340,345, one first and second coding unit 350 of one first and second quantifying unit, 355, and a stream output unit 390.

A/D converter 305 is sampled to input audio signal with the sample frequency (this frequency is the twice of the sample frequency of prior art 1) of 44.1kHz, convert simulated audio signal to digital audio and video signals (for example 16 bits), and produced the voice data string in the time domain.

Voice data input block 310, sample frequency (approximately 45.4msec) with the voice data string (2 frame) that receives 2048 samplings that A/D converter 305 produced, it is exactly the twice of common low sample frequency, the separating audio serial data, it is divided into each the voice data string with 2048 continuous samplings, have two groups of 1024 samplings that obtained before and after 1024 samplings that are capped, be exactly the twice of common sampling number (4096 samplings).Voice data input block 310 comprises a counter 311, is used to detect the separation timing of 2048 samplings of every reception, and a FIF0 buffer, is used for temporarily storing the voice data string of 4096 samplings.

The audio sampling data of 4096 samplings of two frames of the time domain that converter unit 320 is separated voice data input block 310 is transformed into the frequency spectrum data of frequency field.Converter unit 320 comprises a MDCT321, is used for the voice data of 4096 samplings of time domain is transformed into the frequency spectrum data of 4096 samplings of frequency field, and a grouped element 322, is used to divide into groups to be used for the frequency spectrum data of each scale factor.

Detailed says, the sampled data that MDCT321 forms 4096 in time domain samplings is transformed into the frequency spectrum data (16 bit) that comprises 4096 samplings.The arrangement of these sampling quilt symmetries of this frequency spectrum data, so have only the sampling (that is, 2048 samplings) of half to be encoded in the middle of them, and other half is left in the basket.

As mentioned above, if with the A/D converter in the encoding device 300 305, corresponding units in the encoding device 1000 of the structure of voice data input block 310 and converter unit 320 and prior art 1 compares, the essential distinction of present embodiment and prior art 1 is that the sample frequency in the A/D converter 305 is (44.1kHz) that doubles, the separation length of voice data input block 310 is (4096 samplings) that double, and the coding unit among the MDCT321 of converter unit 320 is (4096 sampling) that doubles.

And, if present embodiment is compared with prior art 2, the essential distinction of the former with the latter is that the separation length in voice data input block 310 is that (4096 samplings) that double and the coding unit among the MDCT321 in the converter unit 320 are (4096 samplings) that doubles, although the sample frequency in the A/D converter is identical.

The result is, converter unit 320 output belongs to 1024 samplings (being called as " frequency spectrum data of low-frequency band " later on) of the frequency spectrum data in 11.025kHz or the littler low-frequency band, and 1024 samplings (being called as " frequency spectrum data in the high frequency band " later on) that belong to the frequency spectrum data in the high frequency band that is higher than 11.025kHz, be exactly 2048 frequency spectrum data samplings altogether.

The grouped element 322 of converter unit 320 will be to be encoded 2048 samplings of frequency spectrum data be grouped into a plurality of scale factor, each comprises all that at least the frequency spectrum data that is made of a sampling (perhaps, actual says that the number altogether of sampling is 4 multiple).

According to AAC, the number that is included in the sampling of the frequency spectrum data in each scale factor is defined according to its frequency.The scale factor of low-frequency band is determined narrowlyer by less frequency spectrum data, the scale factor of high frequency band by more frequency spectrum data be determined broad.In AAC, also can define according to sample frequency with the number of the corresponding scale factor of frequency spectrum data of a frame.For example when sample frequency for 44.1kHz for example the time, each frame comprises 49 scale factor, and 49 scale factor comprise the frequency spectrum data of 1024 samplings.On the other hand, it is not specifically defined in AAC, and scale factor will be transmitted in these scale factor, and from the very first reportedly the scale factor of the expectation selected of the transfer rate of defeated channel will be transmitted.When transfer rate for example is 96kbps, only there are 40 scale factor (640 samplings) selectively to be transmitted in the low-frequency band in a frame.

On the other hand, in the present embodiment, the frequency spectrum data in two frames (1024 frequency spectrum datas in low-frequency band and high frequency band respectively) is exported from MDCT321 with the sample frequency (approximately 45.4msec) that doubles in the classic method.Like this, when the transfer rate of transmission channel is 96kbps, if even all proportions factor band in the low-frequency band among two frames (1024 samplings) all is transmitted, in transmission channel, still exist remaining enough capacity, this be with corresponding to the transmission of two frames (640 * 2=1280 sampling) of traditional AAC comparatively speaking.Therefore, present embodiment will suppose that grouped element 332 is grouped into it with the frequency spectrum data of conversion and determines and number is made an explanation by well-determined scale factor.

Data division unit 330 is divided into 1024 frequency spectrum datas in the low-frequency band and 1024 frequency spectrum datas in the high frequency band with 2048 frequency spectrum datas samplings of converter unit 320 output.Data division unit 330 outputs to first quantifying unit 340 with 1024 frequency spectrum datas dividing in the low-frequency band respectively, and 1024 frequency spectrum datas in the high frequency band are outputed to second quantifying unit 345.

First quantifying unit 340 is determined a scale factor that is used for the frequency spectrum data that transmitted from data division unit 330 for each scale factor in the low-frequency band, utilize this scale factor of determining to quantize the interior frequency spectrum of this scale factor, and will be as the quantized value of quantized result, first scale factor of determining, and first and each ensuing scale factor between difference export to first coding unit 350.First quantifying unit 340 comprises a scale factor calculation unit 341.Scale factor calculation unit 341 calculates a normalization factor (scale factor, 8 bits) thus make frequency spectrum data in each scale factor all in the bit of a predetermined number, utilize each frequency spectrum in the scale factor quantization scaling factor frequency band of this calculating, calculate poor between this scale factor and first scale factor then.

The data that first coding unit 350 is quantized first quantifying unit 340, be used for the scale factor of each scale factor etc., be encoded into a predetermined stream format, and comprise that one is used for further compressing each quantized data, the huffman coding table 351 of each scale factor etc.Especially, first coding unit 350 utilizes the data of huffman table 351 each quantification of coding, each scale factor etc., thus can be transmitted with low bit rate.

Second quantifying unit 345 is higher than in the high frequency band of 11.025kHz in the bandwidth that is not quantized by first quantifying unit 340 exactly, and the frequency spectrum data of exporting according to data division unit 330 calculates sub-information, and with its output.Second quantifying unit 345 comprises a sub-information generation unit 346, is used to produce sub-information.

Sub-information is a simplification information of calculating according to the frequency spectrum data in the high frequency band, and the simple and clear characteristic that has shown the frequency spectrum data in the high frequency band with lesser amt information.In other words, it is the information that shows the frequency spectrum data characteristic of high frequency band, and wherein these characteristics are to obtain by the voice data that conversion receives in a specific time span.Particularly, sub-information is the scale factor of each scale factor in the high frequency band of quantized value " 1 " of the frequency spectrum data that is used to obtain the absolute value maximum (absolute value is maximum frequency spectrum data), with and quantized value.

Second coding unit 355 is encoded to predetermined stream format with the sub-information of second quantifying unit, 345 outputs, and the information of output encoder is as second coded message.Second coding unit 355 comprises the huffman coding table 356 of the sub-information that is used to encode.

Stream output unit 390 adds the sub-information of header (header) information and other necessity in first coded signal of above-mentioned first coding unit, 350 outputs, and it is transformed into a MPEG-2 ACC bit stream.Second coded signal that stream output unit 390 is also exported second coding unit 355 is recorded in above-mentioned zone that ignored by traditional decoding device or the bit stream that its operation is not defined.Particularly, the coded signal that stream output unit 390 is exported second coding unit 355 is stored in the filling part (FillElement) of MPEG-2 ACC coded bit stream, the data stream part (Data Stream Element, etc.).

For the information that is stored in the sample frequency that shows bit stream in the header information, the value of half of the sample frequency of voice data is stored.In other words, when the sample frequency of voice data is 44.1kHz, the information of 22.05kHz, promptly half of actual value is stored.And the information that shows the actual samples frequency of 44.1kHz is stored in the above-mentioned stored zone of sub-information or similarly regional.

The bit stream of encoding device 300 outputs utilizes radiowave by transmission medium, optical cable, and flash of light, metal wires etc. are sent to decoding device 400 such as the internet.

As mentioned above, when in the frequency spectrum data that frequency field quantizes and coding is obtained by converter unit 320, encoding device 300 is divided into frequency spectrum data (1024 samplings) in frequency spectrum data (1024 samplings) in the low-frequency band and the high frequency band with it, with traditional method quantification and the frequency spectrum data in the coding low-frequency band, utilize diverse ways to quantize and the interior frequency spectrum data (producing sub-information and the sub-information of encoding) of coding high frequency band, coded bit stream in the high frequency band is synthesized in the low-frequency band, and with its output.Encoding device 300 is that with the essential distinction of traditional encoding device 1000 method that encoding device 1000 is sampled same on the whole quantizes and the coding frequency spectrum data.

The result is that the high-quality sound of regenerating thereby voice data can be encoded can not increase informational capacity simultaneously.

And owing to show that the information of the sample frequency of 22.05kHz is stored in the middle of the header, the effect of generation is that the bit stream that the encoding device 300 in the present embodiment is produced also can utilize traditional decoding device 2000 to decode.

(decoding device 400)

The decoding device in the present embodiment 400 is a kind ofly can carry out that thereby handle can be at the equipment (regeneration frequency is 22.05kHz or littler) of time domain reproducing audio signal by the bit stream of coding unit 300 being exported in the mode approximate opposite with encoding device 300.Decoding device 400 comprises stream input block 410, the first and second decoding units 420,425,430,435, de-quantization data integration unit 440, the first and second de-quantization unit, 480, one voice data output units 490 of an inverse transformation block, and a D/A converter 495.

When receiving encoding device 300 bitstream encoded by transmission medium when, stream input block 410 select one be stored in by first coded signal in the employed zone of traditional decoding device and one be stored in ignored by traditional decoding device or zone that its operation is not defined in second coded signal, and output it to first decoding unit 420 and second decoding unit 425 respectively.

First decoding unit 420 receives first coded signal of stream input block 410 outputs, thereby then its decoding is regenerated as quantized data, and this unit also comprises a huffman table decoder 421.

The quantized data and the output spectrum data of 430 de-quantizations, first decoding unit, 420 decodings of the first de-quantization unit, this unit also comprises a processing unit 431, is used for according to formula the data that quantize being carried out de-quantization.Here, the sampling number of the frequency spectrum data that the first de-quantization unit 430 is exported is 1024, their expression 11.025kHz or littler regeneration bandwidth.

Second decoding unit 425 receives second coded signal that stream input blocks 410 are exported and separates numeral information, and this unit also comprises a huffman table decoder 426.

The second de-quantization unit 435 produces frequency spectrum data in high frequency band, this unit also comprises a frequency spectrum data generation unit 436.Here, the sampling number of the frequency spectrum data that the second de-quantization unit 435 is exported is 1024, and their expressions are higher than the regeneration bandwidth of 11.025kHz.

Frequency spectrum data generation unit 436 is according to predetermined handling procedure, the frequency spectrum data of being exported based on the first de-quantization unit 430 produces noise, the sub-information of being exported according to second decoding unit 425 formalizes to noise, and the frequency spectrum data in the output high frequency band.This noise comprises white noise, the duplicating of the some or all of frequency spectrum data in powder noise (pink noise) and the low-frequency band.

Particularly, frequency spectrum data generation unit 436 duplicates the frequency spectrum data in the low-frequency band of being exported the first de-quantization unit 430 in advance, it is copied in the high frequency band, reconstructed spectrum data by each frequency spectrum data in this scale factor being multiply by and in high frequency band then, this ratio as the ratio of a coefficient be the frequency spectrum data that duplicates in each band in high frequency band bare maximum with to utilize and sub-information in the corresponding scale factor value of band described come the value " 1 " of quantification is carried out ratio between the resulting value of de-quantization.

The frequency spectrum data that the frequency spectrum data that the first de-quantization unit 430 is exported in de-quantization data integration unit 440 and the second de-quantization unit 435 are exported carries out integrated.Here, the sampling number of the frequency spectrum data that de-quantization data integration unit 440 is exported is 2048, and they represent the regeneration bandwidth of 0～22.05kHz.

As mentioned above, decoding device 400 encoding device 300 coded bit streams are divided into first coded signal (in low-frequency band) that is stored in respectively in the employed zone of traditional decoding device and be stored in ignored by traditional decoding device or zone that its operation is not defined in second coded signal (in high frequency band), the same method of utilization and classic method is only decoded and de-quantization to first coded signal (in the low-frequency band), utilize with traditional method diverse ways second coded signal (in the high frequency band) is decoded and de-quantization, be integrated in the frequency spectrum data in high frequency band and the low-frequency band, and export integrated data.In this, the essential distinction of the decoding device 2000 in decoding device 400 and the prior art 1,2 is that decoding device 2000 decodes and de-quantization to the bit stream in whole bandwidth with same method.

The result is, using can decode with the information of traditional approximately uniform lesser amt of method is far longer than the quantity of information of classic method, thus the sound signal high-quality sound of can decodedly regenerating like this.

The frequency spectrum data of the frequency field that 480 pairs of de-quantization data integrations of inverse transformation block unit 440 is exported is carried out IMDCT, and it is transformed into the voice data of 2048 samplings (2 frame) in the time domain.

Voice data output unit 490 merges the voice data of some groups of 2048 samplings of the time domain of inverse transformation block 480 acquisitions mutually, and with time sequencing it is exported one by one.

D/A converter 495 utilizes the sample frequency of 44.1kHz to convert digital audio-frequency data to simulated audio signal.

As mentioned above, the essential distinction of the decoding device 2000 in decoding device 400 and the prior art 1 is that the inverse transformation block in inverse transformation block 480 is (2048 sampling) that doubles, and the frame length in the voice data output unit 490 is that (2048 samplings) that double and the sample frequency in the D/A converter 495 are (44.1kHz) that doubles.

The result is, according to frequency spectrum data (1024 samplings) in 11.024kHz or the lower low-frequency band and the frequency spectrum data (1024 samplings) in the high frequency band, thereby sound signal is output in high frequency band (0～22.05kHz) the high-quality sound of regenerating.

As mentioned above, according to functional structure of the present invention, the high-quality sound thereby voice data can decodedly be regenerated, this is based on and traditional method information content much at one, by with the data in traditional method decoded low frequency band and utilize the frequency spectrum data of considerably less quantity of information decoding high frequency band to realize.

And, in the encoding device 300 and decoding device 400 of present embodiment, data division unit 330, second quantifying unit 345 and second coding unit 355 are to be added in traditional encoding device 1000, second decoding unit, 425, the second de-quantization unit 435 and de-quantization data integration unit 440 have been added to traditional decoding device 2000.Be exactly that encoding device 300 in the present embodiment and decoding device 400 can be under the situations that does not change traditional encoding device 1000 and decoding device 2000 fully and realize with regard to having produced an effect like this.

Also having had another effect is exactly that the bit stream that produced of the encoding device 300 in the present embodiment can be decoded by traditional decoding device 2000.

Next, with the performed encoding process in each unit in the encoding device 300 in the detailed description broadcast system 1.

Accompanying drawing 3A and accompanying drawing 3B are the figure that shows the state variation of the voice data input block 310 of encoding device shown in Figure 2 300 and converter unit 320 handled sound signals.Especially, accompanying drawing 3A shows the waveform of 2048 sampled datas of the time domain that voice data input block 310 shown in Figure 2 is separated, accompanying drawing 3B show the sampled data of time domain by after the MDCT321 conversion in the converter unit shown in Figure 2 320 at the waveform of the frequency spectrum data that frequency field produced.It should be noted that the waveform that the sampled data drawn and frequency spectrum data all are illustrated as simulation in accompanying drawing 3A and 3B, although in fact they all are digital signals.They are real in the figure that shows waveform that fetches.

Voice data input block 310 is received in the voice data of the sample frequency sampling of 44.1kHz.According to this digital audio and video signals, voice data input block 310 is separated into each 2048 the continuous sampling with two groups of 1024 samplings that obtain with voice data before and after 2048 samplings are capped, and these samplings are output to converter unit 320 then.

320 pairs of converter units, 4096 sampled datas are altogether carried out MDCT., therefore only had with corresponding half the frequency spectrum data of 2048 samplings to be output, shown in Fig. 3 B by symmetric offset spread according to the waveform of the frequency spectrum data that MDCT produced.

In accompanying drawing 3B, Z-axis is represented the value of frequency spectrum data, is exactly at 2048 points corresponding to sampling number, the quantity (size) of the frequency component of the voice data of representing at the magnitude of voltage of 2048 samplings of expression shown in the accompanying drawing 3A.Owing to be input to the sound signal of encoding device 300 is to carry out the A/D conversion with the sample frequency of 44.1kHz, so the regeneration bandwidth of frequency spectrum data is 22.05kHz.And because the frequency spectrum that MDCT321 produced has the negative value shown in Fig. 3 B, in the time of the coding frequency spectrum, the sign symbol of the frequency spectrum that MDCT321 produced also needs to be encoded.In the middle of the explanation below, the information of the sign symbol of expression frequency spectrum data is called as " symbolic information ".

Frequency spectrum data that converter unit 320 is exported and symbolic information be divided into by data division unit 330 0～11.025kHz low-frequency band data and information and be higher than data and information in the high frequency band of 11.025kHz, frequency spectrum data and symbolic information in the low-frequency band are output to first quantifying unit 340, and data and information in the high frequency band are output to second quantifying unit 345.

Accompanying drawing 4 is to show that first quantifying unit, 340 performed scale factors shown in Figure 2 determine the operational flowchart of handling.

First quantifying unit 340 determines that at first one is the common scale factor initial value (S91) as scale factor for each scale factor, utilize predetermined scale factor to quantize interior all frequency spectrum datas that will be transmitted of low-frequency band as a frame voice data (1024 samplings), calculating poor before the scale factor that calculates and between the scale factor afterwards carries out huffman coding (S92) to the quantized value of poor, first scale factor and frequency spectrum data.It should be noted that here quantification and coding only are performed for the count bits number.Therefore, only there are data to be quantized and encode, for the information of simplifying processing such as header is not added into.

Next, first quantifying unit 340 judges whether the bit number of huffman code data has surpassed a predetermined bit number (S93), if exceeded, reduces the initial value (S101) of scale factor.Then, the scale factor value that the utilization of first quantifying unit 340 reduces quantize once more with the huffman coding low-frequency band in same frequency spectrum data (S92), whether the bit number of judging the data of the huffman coding that is used for a frame in the low-frequency band has exceeded predetermined bit number (S93), and repeats this process and become the bit of predetermined number or still less until it.

The bit number of the coded data in low-frequency band does not exceed predetermined value, and first quantifying unit 340 repeats following processing procedure to each scale factor, and has determined the scale factor (S94) of each scale factor.At first, the value de-quantization (S95) of each quantification in the Comparative Examples factor band, the value that computational solution quantizes is poor with the absolute value between the raw spectroscopy data value accordingly, and with its addition (S96).Next, the summation of the difference of judge calculating whether in the acceptable limit with interior (S97), if it in acceptable limit, repeats above-mentioned processing procedure (S94～S98) in next scale factor.

On the other hand, it has exceeded receivable limit, first quantifying unit 340 increases scale factor value, quantize the frequency spectrum data (S100) in this scale factor, next the value (S95) that quantizes of de-quantization and with the absolute value of the value of de-quantization and the poor addition (S96) between the corresponding frequency spectrum data value.Next, the summation that first quantifying unit 340 is judged difference whether in receivable limit with interior (S97), if it has exceeded limit (S100), just increase scale factor value and become a value (S100) that is positioned at limit, and repeat above-mentioned processing procedure (S95～S97 and S100) until it.

When first quantifying unit 340 is judged for all scale factor, in the summation of the difference of the data that quantize being carried out data value that de-quantization obtains and the absolute value between the corresponding raw spectroscopy data value with scale factor in acceptable limit with interior (S98), it uses the scale factor of determining once more the frequency spectrum data in low-frequency band of a frame to be quantized, poor to each scale factor, the quantized value of first scale factor and this frequency spectrum data carries out huffman coding, and judges whether the bit number of the coded data in the low-frequency band has exceeded a predetermined bit number (S99).If the bit number of the coded data in the low-frequency band has exceeded predetermined value, first quantifying unit 340 reduces the initial value of scale factor, become predetermined number or littler (S101) until it, in each scale factor, repeat to determine the processing procedure (S94～S98) of scale factor then.If the bit number of the coded data in the low-frequency band does not exceed predetermined value (S99), so just determine that the value of each scale factor at that time is the scale factor that is used for each scale factor.

The scale factor that first quantifying unit 340 is determined above utilizing quantizes the frequency spectrum data in the low-frequency band, and the value that will quantize, first scale factor, poor between first scale factor of determining and the ensuing scale factor, and the symbolic information that data division unit 330 is received is exported to first coding unit 350.

It should be noted that in scale factor to the data that quantize carry out data value that de-quantization obtains and the absolute value between the corresponding raw spectroscopy data value difference summation whether acceptable limit with interior be to carry out estimative according to the data of acoustic mode etc.

And, in above-mentioned situation, relatively large value is set the initial value as scale factor, when the bit number of the huffman coding in the low-frequency band exceeds a predetermined bit number, thereby the initial value of scale factor just is reduced to determine scale factor, but scale factor is not often need be determined by this way.For example, a lower value can be redefined for the initial value of scale factor, and the increase that initial value can be gradually.And the scale factor in each scale factor can the proportion of utilization factor initial value be determined, this scale factor initial value is to be set before the bit number altogether of the coded data in low-frequency band exceeds a predetermined bit number first.

And in the present embodiment, the bit number altogether that the scale factor of each scale factor all is determined the coded data of a frame that makes that low-frequency band is interior is no more than predetermined number, but scale factor is not always to need to be determined by this way.For example, thus scale factor can be determined and makes that in each scale factor the value of each quantification in scale factor does not exceed predetermined bit number.The operating process of 5 explanations, first quantifying unit 340 in this processing procedure below with reference to accompanying drawings.

Accompanying drawing 5 is the definite operational flowcharts handled of another scale factor that show first quantifying unit 340 shown in Figure 2.

First quantifying unit 340 is for all scale factor, the scale factor (S1) in the low-frequency band that calculating will be encoded according to following processing procedure.And first quantifying unit 340 is calculated the scale factor (S2) that is used for all frequency spectrum datas in each scale factor according to following processing procedure.

At first, first quantifying unit 340 utilizes a predetermined scale factor value to quantize frequency spectrum data (S3) according to formula, and judges whether the value that quantizes surpasses the given predetermined bit number that is used to show quantized value, for example 4 bits (S4).

When the value that quantizes exceeded 4 bits as judged result, first quantifying unit 340 was with regard to the value (S8) of the resize ratio factor, and the scale factor value that utilization is adjusted quantizes same frequency spectrum data (S3).First quantifying unit 340 judges whether the quantized value that obtains exceeds 4 bits (S4), and repeats the adjustment (S8) of scale factor and the quantification (S3) of the scale factor adjusted is 4 bits or still less until the quantized value of frequency spectrum data.

When the value that quantizes is 4 bits or still less as judged result the time, it will utilize the scale factor value of being scheduled to quantize next frequency spectrum data.

When the value of the quantification of all frequency spectrum datas in the scale factor is 4 bits or still less the time (S5), first quantifying unit 340 just determines that scale factor at that time is the scale factor (S6) that is used for scale factor.

After the scale factor of having determined all proportions factor band (S7), first quantifying unit 340 just finishes this processing.

According to top processing, with the scale factor in the low-frequency band that is encoded, each scale factor all is determined for all.The scale factor that 340 utilizations of first quantifying unit are determined in above-mentioned mode quantizes the frequency spectrum data in the low-frequency band, and will be as the quantized value of 4 bits of quantized result, first scale factor of 8 bits, poor between the scale factor of first scale factor and back also has the symbolic information that receives from data division unit 330 to output to first coding unit 132.

The value that quantizes, other value of scale factor and 350 outputs of first coding unit is a huffman coding, and as in downsampled situation, is output to stream output unit 390 as first coded signal.

On the other hand, second quantifying unit 345 is according to sub-information of generation such as the frequency spectrum datas in the high frequency band.

Accompanying drawing 6 is spectrum waveform figure of the embodiment of the sub-information (scale factor) that shows that second quantifying unit 345 shown in the accompanying drawing 2 is produced.Accompanying drawing 7 is operational flowcharts of sub-information (scale factor) computing that shows that second quantifying unit 345 shown in Figure 2 is performed.

In Fig. 6, the boundary's symbol (delimiter) that marks on the frequency axis of low-frequency band shows the scale factor of the scale factor of determining in the present embodiment.And boundary's symbol that with dashed lines marks on the frequency axis in high frequency band shows the scale factor of the scale factor in the high frequency band of determining in the present embodiment.These are real in ensuing oscillogram.

In the frequency spectrum data of converter unit 320 outputs, 11.025kHz or the littler interior regeneration bandwidth of low-frequency band are represented with solid line waveform in accompanying drawing 6, are output to first quantifying unit 340, and as usual it are quantized.On the other hand, the regeneration bandwidth in being higher than the high frequency band of 11.025kHz～22.05kHz dots in Fig. 6, sub-information (scale factor) representative that is calculated by second quantifying unit 345.

7 process flow diagram below with reference to accompanying drawings utilizes the embodiment shown in the accompanying drawing 6 to explain sub-information (scale factor) computation process in second quantifying unit 345.

According to following processing (S11), second quantifying unit 345 is calculated and is used for deriving the optimal scale factors of quantized value " 1 " of the frequency spectrum data of absolute value maximum in greater than each scale factor of each scale factor of 11.025kHz in the high frequency band of 22.05kHz having the regeneration bandwidth.

Second quantifying unit 345 specifies in to be had greater than the absolute maximum spectrum data (peak value) of first scale factor in the high frequency band of the regeneration bandwidth of 11.025kHz (S12).In the embodiment shown in fig. 6,1. be illustrated in the peak value of appointment in first scale factor, the value of peak value is " 256 ".

According to the processing procedure identical with the process flow diagram shown in Fig. 5, second quantifying unit 345 calculate be used for deriving the quantized value " 1 " that obtains by the initial value of giving peak value " 256 " and scale factor at formula scale factor value " sf " (S13).In this case, for example, " sf "=24 that calculate (" sf " is the scale factor of the quantized value " 1 " that is used to derive peak value " 256 ").

When being that first scale factor is when calculating the scale factor " sf "=24 of the peak value " 1 " that is used to derive quantification (S14), second quantifying unit 345 is specified the peak value (S12) of the frequency spectrum data in the next scale factor, if the peak of appointment be 2. and value for " 312 ", just calculate the value of the scale factor of the quantized value " 1 " that is used to derive peak value " 312 ", for example, " sf "=32 (S13).

In the same way, second quantifying unit 345 calculate respectively being used in the high frequency band derive peak value 3. the quantized value of " 288 " be the scale factor value of the 3rd scale factor of " 1 ", " sf "=26, and be used to derive peak value 4. " 203 " quantized value be the scale factor value of the 4th scale factor of " 1 ", for example " sf "=18.

When the quantized value that is used to derive peak value that has calculated each scale factor in the high frequency band by this way for the scale factor of " 1 " time (S14), second quantifying unit 345 will be calculated the scale factor of each scale factor of gained and export to second quantifying unit 355 as the sub-information in the high frequency band, end process then.

Sub-information (scale factor) produces by second quantifying unit 345, as mentioned above.If for each scale factor in high frequency band (in this case, being 4 frequency bands), the sub-information of representing with 1024 samplings of frequency spectrum data (scale factor) is to represent with 0 to 256 digital value, and it can be represented by enough 8 bits so.And if the difference of corresponding scale factor is a huffman coding, data volume can further reduce probably.On the other hand, if 1024 samplings of the frequency spectrum data in the high frequency band are quantized and huffman coding with the processing mode in traditional low-frequency band, the data predicted amount is at least 300 bits.Therefore, this sub-information only is to have represented a scale factor of each scale factor in the high frequency band, clearly, compares with the quantification of classic method medium-high frequency band, and data volume has significantly reduced.

And, scale factor represented in each scale factor with peak value (absolute value) approximately proportional value, therefore we can say in the high frequency band for the frequency spectrum data of 1024 samplings of fixed value or the frequency spectrum data reconstruction roughly that obtained by the copy of the some or all of frequency spectrum data in the low-frequency band and scale factor are multiplied each other according to frequency spectrum data that input audio signal obtained.And, for each scale factor, frequency spectrum data can rebuild more accurately by the ratio that each frequency spectrum data in the frequency band be multiply by as a coefficient, and this ratio is with the bare maximum of the frequency spectrum data of this spectral band replication and utilizes the ratio that quantized value " 1 " is carried out the resulting value of de-quantization with the corresponding scale factor value of this frequency band.And the difference of the waveform in the high frequency band does not have the obvious difference in the low-frequency band, and therefore the sub-information of above-mentioned acquisition is enough as the information of representing the waveform in the high frequency band.

In the present embodiment, be " 1 ", but not need always be " 1 " for it, can be other value thereby scale factor calculates the quantized value of the frequency spectrum data in each scale factor that makes in the high frequency band.

The sub-information that second quantifying unit 345 produces is by second coding unit, 355 huffman codings, and is stored in the zone of that traditional decoding device is ignored or the bit stream that its operation is not defined as second coded signal by stream output unit 390.

Accompanying drawing 8A～8C is the figure in zone of bit stream that shows the sub-information of stream output unit shown in Figure 2 390 storage.In these figure, the sub-information of the frequency spectrum in the expression high frequency band is encoded, and is stored in the zone that is not identified as second audio coding signal in the bit stream as second coded signal then.

In accompanying drawing 8A, dash area is a zone that is called as filling part (fill element), and it is received in " 0 " thus the data length of unification bit stream.Even the sub-information of the frequency spectrum in the expression high frequency band is exactly second coded signal, be stored in this zone, in traditional decoding device 2000, it can not be identified as decoded coded signal is left in the basket.

In accompanying drawing 8B, for example dash area is the zone that is called as data stream part (DSE).Pre-interim being provided that this zone will be expanded in the future of MPEG-2 AAC, and have only its physical arrangement in MPEG-2 AAC, to be defined.As in filling part, even the sub-information of the frequency spectrum in the expression high frequency band is stored in this zone, traditional decoding device 2000 is ignored it, and perhaps owing to should be defined by the operation that traditional decoding device 2000 is carried out, it can not carry out any operation in response to reading information.

In the superincumbent description, be included in second coded signal in the MPEG-2 AAC bit stream and be stored in the zone of being ignored by traditional decoding device 2000.But second coded signal can be integrated in the predetermined zone of header information, perhaps in the presumptive area of first coded signal, perhaps in the two the zone of the header and first coded signal.For storage second coded signal in bit stream does not need to guarantee the continuum of the header information and first coded signal.For example, second coded signal can directly be integrated between the header information and first coded message, shown in Fig. 8 C.

Accompanying drawing 9A and 9B show the figure of other embodiment in zone of bit stream of the sub-information of stream output unit shown in Figure 2 390 storage.Accompanying drawing 9A shows a stream 1, in this stream, only has first coded signal to be stored continuously in each frame.Accompanying drawing 9B shows a stream 2, and second coded signal is only arranged in each frame in this stream, and the sub-information that is exactly coding is by storage is corresponding with stream 1 continuously.

Second coded signal of stream output unit 390 in can storage flow 2, stream 2 is different fully with the stream 1 of wherein storing first coded signal.For example, stream 1 is the bit streams that transmit by different channels with stream 2.

As mentioned above, because the low-frequency band of the essential information of expression input audio signal is transmitted in advance or is stored by transmit first and second coded signals in diverse bit stream, just produced an effect,, can add the information of high frequency band in the back if necessary.

In the form shown in accompanying drawing 8A and 8B and accompanying drawing 9A and the 9B, show it is that half the information of 22.05kHz of actual samples frequency is stored in and shows and will be stored in the middle of the information of sample frequency of the bit stream in the header.Therefore, though the decoding device 2000 in the prior art 1 to also can separate under the situation of down-sampling in the frequency band of code frequency at 0～11.025kHz bit stream and it can regenerated.

The difference of the method for the method of the encoding device 300 of present embodiment and the encoding device 1000 of prior art 1 10A and 10B with reference to the accompanying drawings makes an explanation.Accompanying drawing 10A and 10B show the comparison between the method for present embodiment and prior art 1.Particularly, accompanying drawing 10A has shown the method in the present embodiment, and accompanying drawing 10B has shown the method in the prior art 1.

Method according to present embodiment, obtain a string voice data string in the sample frequency of 44.1kHz second every 22.7 μ, 4096 sampled datas altogether, be exactly, be included in two groups of 1024 samplings before and after in the frame that will be encoded 2048 samplings and this frame, these data are separated and be performed MDCT, just obtained 2048 samplings of frequency spectrum data then.The regeneration bandwidth of this frequency spectrum data is represented 22.05kHz.2048 samplings of these of frequency spectrum data are that boundary is divided into frequency spectrum data (1024 samplings) and the interior frequency spectrum data (1024 samplings) of high frequency band in the low-frequency band with 11.025kHz.Frequency spectrum data (1024 samplings) in the low-frequency band is quantized in common mode and encodes, and has so just obtained as downsampled first coded signal better quality and low bit rate that has.And the frequency spectrum data of 1024 samplings of high frequency band is also obtained.If these data are quantized in common mode and encode, just can't realize low bit rate.Therefore, in the method for present embodiment, the sub-information of 1024 sampling generations according to the frequency spectrum data in the high frequency band has obtained second coded signal by only this sub-information being encoded.Therefore thereby a sound signal high-quality sound of regenerating that can be encoded can not increase informational capacity simultaneously basically.

On the other hand, prior art 1 in the method for down-sampling, obtain a string voice data string in the sample frequency of 22.05kHz second every 45 μ, 2048 sampled datas altogether, take a sample and two groups of 512 samplings before and after this frame for 1024 that just are included in the frame that to be encoded, these data are separated and be performed MDCT, just obtained 1024 samplings of frequency spectrum data then.The regeneration bandwidth of this frequency spectrum data is represented 11.025kHz.1024 samplings of frequency spectrum data are quantized in common mode and encode.Like this, just can obtain the high-quality coded signal in 11.025kHz or littler bandwidth, but can't obtain to be higher than the coded signal in the high frequency band of 11.025kHz, because in high frequency band, there is not frequency spectrum data.

Next, 11A and 11B describe difference between the method for encoding device of the method for encoding device 300 of present embodiment and prior art 2 with reference to the accompanying drawings.

Accompanying drawing 11A and 11B have shown the comparison between the method for the sharp and prior art 2 of this enforcement.Particularly, accompanying drawing 11A shows the method in the present embodiment, and accompanying drawing 11B shows the method for prior art 2.Because the method in the present embodiment was explained in the above, will omit relevant description here.

In the method for sampling of prior art 2, to obtain a string voice data string second every 22.7 μ in the sample frequency of 44.1kHz, 2048 sampled datas altogether, take a sample and two groups of 512 samplings before and after this frame for 1024 that just are included in the frame that to be encoded, these data are separated and be performed MDCT, just obtained 1024 samplings of frequency spectrum data then.The regeneration bandwidth of this frequency spectrum data is represented 22.05kHz.1024 samplings of frequency spectrum data are quantized in common mode and encode.In other words, the time of half of the time span in present embodiment (22.7msec) has obtained 1024 samplings (512 samplings in the high frequency band of 512 11.025kHz that take a sample and be higher than in 11.025kHz or the lower low-frequency band) of frequency spectrum data.

Here, suppose in the encoding device 1000 of prior art 2,, produce sub-information certainly the interior frequency spectrum data of high frequency band that is higher than 11.025 to 22.05kHz as the situation in the embodiments of the invention.In this case, when being when " n " and the bit number that can be used as sub-information are " m1 " at the bit number that uses in the quantification of carrying out every 22.7msec, 512 samplings in the low-frequency band (0-11.025kHz) need to utilize (n-m1) individual bit to quantize.On the other hand, in the present embodiment, when being when " 2xn " and the bit number that can be used as sub-information are " m2 " at the bit number that uses in the quantification of carrying out every 45.4msec, (the individual bit of 1024 samplings available (2xn-m2) in 0～11.025kHz) quantizes low-frequency band.

By the way, as everyone knows,,, just can't realize high code efficiency if can not obtain some or more sampling according to AAC.512 samplings in the prior art 2 do not reach threshold value, and 1024 samplings in the present embodiment have exceeded this threshold value greatly.

Therefore,, quantize if 1024 samplings are utilized (2xn-m2) individual bit, rather than as prior art 2, utilize (n-m1) individual sampling that 512 samplings are quantized, just can realize higher code efficiency according to present embodiment.And owing to realized higher code efficiency in the present embodiment, " m2 " can be bigger (m2＞2 * x1), so the sound quality in the high frequency band can be enhanced.

Accompanying drawing 12 has shown the frequency spectrum data in the coding method of present embodiment and prior art 1 and 2 and the comparison of characteristic.

In the present embodiment, sample frequency is 44.1kHz, and frame length is 2048 samplings.Therefore, just can obtain in the low-frequency band of 0～11.025kHz 1024 of frequency spectrum data samplings and based on the sub-information of 1024 frequency spectrum datas in the high frequency band.The result is, it is identical that bandwidth and bandwidth in the prior art 2 are close to, but wideer than the bandwidth of prior art 1.And the sound quality in the low-frequency band of 0～11.025kHz is identical with prior art 1, but the sound quality in being higher than the high frequency band of 11.025kHz is higher as a complete unit, because have sub-information in the present embodiment.In addition, owing to have sub-information, present embodiment is being higher than the identical of the sound quality of 11.025kHz in the high frequency band of 22.05kHz and prior art 2, but the sound quality in the low-frequency band of 0～11.025kHz is higher, because the number of frequency spectrum data is doubled.Like this, generally speaking, the sound quality in the present embodiment is higher.

On the other hand, in prior art 1, sample frequency is 22.05kHz, and frame length is 1024 samplings.1024 samplings of frequency spectrum data obtain in the low-frequency band of 0～11.025kHz.The result is, the bandwidth of prior art 1 is narrower only to be half of present embodiment.Therefore, identical in the sound quality in the low-frequency band of 0～11.025kHz and the present embodiment still is lower than present embodiment to the sound quality in the high frequency band of 22.05kHz being higher than 11.025kHz, because there is not frequency spectrum data in the prior art 1 in high frequency band.Therefore, as a complete unit, the sound quality of prior art 1 is lower.

And in prior art 2, sample frequency is 44.1kHz, and frame length is 1024 samplings.1024 samplings of frequency spectrum data are to obtain in the whole frequency band of 0～22.05kHz.The result is, the bandwidth of prior art 2 is identical with present embodiment, but because the number of frequency spectrum data is reduced to half, the sound quality in the low-frequency band of 0～11.025kHz is lower than present embodiment by deterioration.Although because frequency spectrum data is encoded, it is higher than present embodiment being higher than 11.025kHz in the high frequency band of 22.05kHz.But as a complete unit, the sound quality of prior art 2 is lower.

Therefore, according to present embodiment, by the data in the low-frequency band of encoding in common mode and utilize data in the very small amount of information coding high frequency band, the high-quality sound of a regenerating thereby sound signal just can be encoded, and can not increase the total amount of information.

Next, with the decoding processing of detailed each unit that is described in the decoding device 400 in the broadcast system 1.

Be decoded into quantized data etc. from first coded signal of stream input block 410 output by first decoding unit 420, and be encoded into frequency spectrum data in the low-frequency band by the first de-quantization unit 430.On the other hand, second decoded signal of stream input block 410 outputs is decoded into sub-information by second decoding unit 425.The second de-quantization unit 435 is according to the frequency spectrum data in this sub-information generation high frequency band.Next will describe the processing procedure of the second de-quantization unit 435 in detail.

Accompanying drawing 13 be one show the second de-quantization unit 435 shown in the accompanying drawing 2 with the frequency spectrum of 1024 samplings in the low-frequency band along the process flow diagram that is copied to the processing procedure of high frequency band on the forward direction (forward direction).When the frequency spectrum data in the high frequency band produced, the frequency spectrum data in the low-frequency band was replicated.

In accompanying drawing 13, inv_spec1[i] value of i the frequency spectrum of expression from the data of the first de-quantization unit, 430 outputs, inv_spec2[j] value of j the frequency spectrum of expression from the data that the second de-quantization unit 435 is imported.

At first, the initial value that counter i sum counter j is set in the second de-quantization unit 435 is " 0 ", and the number of rolling counters forward frequency spectrum data, thereby on same direction, import from 0 to the 1023rd frequency spectrum data (S71).Next, less than " 1024 " (S72) whether the value of the second de-quantization unit, 435 inspection counter i.When the value of counter i less than " 1024 " time, the value of i (being the 0th in this case) frequency spectrum data in the low-frequency band of 435 inputs, the first de-quantization unit 430, the second de-quantization unit is as the value (S73) of the j in the high frequency band of the second de-quantization unit 435 (being the 0th in this case) frequency spectrum data.Then, the second de-quantization unit 435 adds 1 (S74) with the value of counter i sum counter j respectively, and less than " 1024 " (S72) whether the value of inspection counter i.

When the value of counter i less than " 1024 " time, the second de-quantization unit 435 repeats above-mentioned process, when its value becomes " 1024 " or bigger the time, this processing procedure finishes.

The result is, is duplicated into the frequency spectrum data in the high frequency band of the second de-quantization unit 435 as the 0th～1023 all frequency spectrum datas in the de-quantization result's of the first de-quantization unit 430 the low-frequency band.

Amplitude according to the frequency spectrum data of the sub-information reproduction of second decoding unit 425 decoding, be exactly, be used to derive the value of the scale factor of peak value " 1 ", be adjusted, and controlled frequency spectrum data is output as the frequency spectrum data of high frequency band.Amplitude is to adjust by each frequency spectrum data in the frequency band be multiply by a ratio, this ratio is the bare maximum of the replica spectra data in this frequency band and value by use and the corresponding scale factor of this frequency band is carried out ratio between the value that de-quantization obtains to quantized value " 1 ", and this ratio is as a coefficient of each scale factor.Here, the maximum number of the sampled data of the frequency spectrum data of the second de-quantization unit, 435 outputs is 1024, and their expressions are higher than the regeneration bandwidth of 11.025kHz.

Be used for 1024 frequency spectrum datas in the low strap frequency band shown in Figure 13 along the processing procedure that the forward direction of frequency axis copies in the high frequency band, but they also can be replicated in the opposite direction, as shown in figure 14.

Accompanying drawing 14 is second a de-quantization unit 435 shown in the expression accompanying drawing 2 copies to 1024 frequency spectrums in the low-frequency band processing procedure in the high frequency band along the reverse direction of frequency axis process flow diagrams.In accompanying drawing 14, as the situation in the accompanying drawing 13, inv_spec1[i] value of i the frequency spectrum data of expression from the data of the first de-quantization unit 430 output, inv_spec2[j] value of j the frequency spectrum data of expression from the data that the second de-quantization unit 435 is imported.

At first, the initial value of the second de-quantization unit, 435 set-up and calculated device i is " 0 ", and the initial value of counter j is " 1023 ", the number of its counting frequency spectrum data, thus import from 0 to the 1023rd frequency spectrum data (S81) in the opposite direction.Next, less than " 1024 " (S82) whether the value of the second de-quantization unit, 435 inspection counter i.When the value of counter i less than " 1024 " time, the value of i (being the 0th in this case) frequency spectrum data in the low-frequency band of 435 inputs, the first de-quantization unit 430, the second de-quantization unit is as the value (S83) of the j in the high frequency band of the second de-quantization unit 435 (being the 0th in this case) frequency spectrum data.Then, the second de-quantization unit 435 adds " 1 " with the value of counter i, and the value of counter j subtracts 1 (S84), and less than " 1024 " (S82) whether the value of inspection counter i.

When the value of counter i less than " 1024 " time, the second de-quantization unit 435 repeats above-mentioned process, when its value becomes " 1024 " or bigger the time, this processing procedure finishes.

The result is, is duplicated into the 0th～1023 frequency spectrum data in the high frequency band of the second de-quantization unit 435 in the opposite direction as the 0th～1023 all frequency spectrum datas in the de-quantization result's of the first de-quantization unit 430 the low-frequency band.

Identical with the front, the amplitude according to the frequency spectrum data of the sub-information reproduction of second decoding unit 425 decoding, be exactly, be used to derive the value of the scale factor of peak value " 1 ", be adjusted, and controlled frequency spectrum data is output as the data of high frequency band.Amplitude is by adjusting in each scale factor each frequency spectrum data of frequency band being multiply by the ratio as coefficient, and this ratio is the bare maximum of the frequency spectrum data that duplicates in the frequency band and by using and the value of the corresponding scale factor of this frequency band is carried out ratio between the value that de-quantization obtains to quantized value " 1 ".Here, the maximum number of the sampled data of the frequency spectrum data of the second de-quantization unit, 435 outputs is 1024, and their expressions are higher than the regeneration bandwidth of 11.025kHz.

In the present embodiment, the second de-quantization unit 435 copies to all frequency spectrum datas in the low-frequency band in the high frequency band, but it also can only duplicate a part wherein.

The example that once all duplicates the processing procedure of high frequency band and low-frequency band has been described with reference to attached Figure 13 and 14.But the part in them can be duplicated according to process shown in Figure 13, and another part duplicates according to process shown in Figure 14.

And, some or all of can being replicated in them by sign symbol reversing with them.

These reproduction processes can be scheduled to, and perhaps can change according to the data in the low-frequency band, perhaps are transmitted as sub-information.

In the present embodiment, the frequency spectrum data in the low-frequency band is replicated as the data in the high frequency band, but the present invention is not limited thereto, and the frequency spectrum data in the high frequency band can only produce according to second coded message.

In the present embodiment, for the generating noise in the second de-quantization unit 435, the situation that the main frequency spectrum data that is obtained by the first de-quantization unit 430 is replicated has been described.But, the present invention is not limited thereto, frequency spectrum data, white noise, the data that have particular value in each scale factor in high frequency band such as pink noise can be produced in the second de-quantization unit 435 in its oneself mode, perhaps produce according to sub-information.

Integrated from 1024 frequency spectrum datas that 1024 samplings of the frequency spectrum data of the second de-quantization unit 435 output and the first de-quantization unit 430 the de-quantization data integration unit 440 are exported, and be performed MDCT, it is transformed into the voice data of time domain, carry out the D/A conversion with the sample frequency of 44.1kHz then, the sound signal that next has the regeneration bandwidth of 0～22.05kHz just is reproduced.

As mentioned above, according to the present invention, 1024 samplings of in the frequency spectrum data of 2048 samplings first use MDCT and IMDC to be encoded in common mode, transform length with traditional approach twice, 1024 samplings of half of back are used than a spot of information of traditional approach coding, thereby two frequency spectrum datas all are integrated and are used for decoding.

The required quantity of information of frequency spectrum data of half 1024 sampling of back can be lowered owing to be used to encode, half the required quantity of information of frequency spectrum data of 1024 samplings of front of being used to encode can be increased, therefore, the frequency spectrum data that is higher than wide bandwidth can be encoded, and the regeneration precision of the original signal in the low-frequency band can be enhanced simultaneously.

And the bit stream that encoding device produced in the present embodiment can be decoded by traditional decoding device.

Next, the various modification of monk information and its decoding will be separated.

Accompanying drawing 15 shows the spectrum waveform of the embodiment of other the sub-information (quantized value) that shows that second quantifying unit 345 shown in Figure 2 is produced.Accompanying drawing 16 is one and shows the operational flowchart in the computing of other performed sub-information (quantized value) of second quantifying unit 345 shown in Fig. 2.

Second quantifying unit 345 has preestablished the value of a scale factor, for example " 18 ", to be higher than 11.025kHz be common to all proportions factor band in the high frequency band of the regeneration bandwidth of 22.05kHz to this scale factor for having, and uses this scale factor value " 18 " to calculate the quantized value (S21) of absolute maximum frequency spectrum data (peak value) in each scale factor.

Absolute maximum spectrum data (peak value) in first scale factor that second quantifying unit 345 is specified in the high frequency band with the regeneration bandwidth that is higher than 11.025kHz (S22).In the embodiment shown in Figure 15,1. be illustrated in the peak value of appointment in first scale factor, peak value at that time is " 256 ".

Second quantifying unit 345 is updated to the formula that is used for calculating quantized value by the common scale factor value " 18 " that will be scheduled to and peak value " 256 " and calculates the value (S23) of quantification.For example, if peak value " 256 " proportion of utilization factor values " 18 " quantizes, the quantized value of calculating is " 6 ".

When being first scale factor bandwidth meter when calculating the quantized value " 6 " of peak value " 256 " (S24), second quantifying unit 345 is just specified the peak value (S22) of the frequency spectrum data in the next scale factor.If 2. the peak of appointment is, and peak value is " 312 ", for example, so proportion of utilization factor values " 18 " calculate peak value " 312 " quantized value " 10 " (S23).

In the same way, second quantifying unit, 345 proportion of utilization factor values " 18 " are calculated the 3. quantized value of " 288 " " 9 " of peak value of the 3rd scale factor of high frequency band, and proportion of utilization factor values " 18 " the 4. quantized value of " 203 " " 5 " of peak value that calculates the 4th scale factor.

When utilizing fixing scale factor value " 18 " to calculate the quantized value of peak value of all proportions factor band in the high frequency band (S24), second quantifying unit 345 will be calculated the quantized value of each scale factor of gained and export to the sub-information of second coding unit 355 as high frequency band, and this processing then is through with.

As mentioned above, second quantifying unit 345 has produced sub-information (quantized value).This sub-information represents with 4 scale factor in the high frequency bands of 1024 of frequency spectrum data sampling expressions with the quantized value of 4 bits respectively, and above-mentioned sub-information (scale factor) is represented 4 scale factor in the high frequency band with the frequency spectrum data of 8 bits respectively.Therefore, under the situation of quantized value, the data volume in the high frequency band reduces greatly.And, the expression roughly of the value of this quantification the amplitude (absolute value) of peak value of each scale factor, also we can say 1024 samplings of frequency spectrum data in the high frequency band of getting fixed value or only multiply by frequency spectrum data reconstruction roughly that quantized value obtains according to frequency spectrum data that input audio signal obtained by copy with the part or all of frequency spectrum data in the low-frequency band.And, for each scale factor, frequency spectrum data can rebuild more accurately by each frequency spectrum data in the frequency band be multiply by the ratio as coefficient, and this ratio is the ratio between the value that obtained of the bare maximum of the frequency spectrum data that duplicates in the frequency band and de-quantization and the corresponding quantized value of this frequency band.

In the present embodiment, preestablished with the corresponding scale factor value of quantized value that will be used as the transmission of second coded message, still best scale factor value can be calculated and be added in second coded message to be transmitted.For example, if it is chosen to be used to obtain the scale factor of maximal value " 7 " of quantized value, the bit number of expression quantized value only be " 3 ", so is used to transmit the required information content of quantized value and will reduces greatly.

Accompanying drawing 17 has shown the spectrum waveform of the example of another sub-information (positional information) of representing that second quantifying unit 345 shown in Figure 2 is produced.Accompanying drawing 18 is process flow diagrams, has shown the operating process of the computing of another sub-information (positional information) that second quantifying unit 345 shown in Figure 2 is performed.

Second quantifying unit 345 is higher than the position (S31) of specifying the frequency spectrum data of absolute value maximum in 11.025kHz each scale factor in the high frequency band of the regeneration bandwidth of 22.05kHz according to ensuing processing procedure having.

The frequency spectrum data (peak value) of the absolute value maximum of appointment (S32) in second quantifying unit 345 first scale factor in the high frequency band with the regeneration bandwidth that is higher than 11.025kHz.In the embodiment shown in Figure 17,1. be illustrated in the peak value of appointment in first scale factor and the 22nd frequency spectrum data that begins from first data of this scale factor.The peak " from first the 22nd frequency spectrum data that begins of this scale factor " of second quantifying unit, 345 these appointments of maintenance (S33).

When for the peak of first scale factor designated and maintained the time (S34), second quantifying unit 345 is just specified the peak value (S32) of the frequency spectrum data in the next scale factor.For example, the peak value of appointment be positioned in 2. and in this frequency band from first the 60th frequency spectrum data (S32) that begins.The peak " from first the 60th frequency spectrum data that begins of this scale factor " of second quantifying unit, 345 these appointments of maintenance (S33).

In the same way, second quantifying unit 345 is specified and is kept peak 3. " first frequency spectrum data of this scale factor " and specify and keep peak 4. " first the 25th frequency spectrum data that begins in this scale factor " in the 4th scale factor in the 3rd scale factor of high frequency band.

All designated and when keeping (S34) when the peak of all proportions factor band in the high frequency band, second quantifying unit 345 outputs to the sub-information of second coding unit 355 as high frequency band with the peak of the scale factor of maintenance, and this processing procedure finishes then.

As mentioned above, second quantifying unit 345 has produced sub-information (positional information).This sub-information (positional information) is represented with 4 scale factor in 1024 high frequency bands of representing of taking a sample of frequency spectrum data with the positional information of 6 bits respectively.

In this case, according to sub-information (positional information) from 425 inputs of second decoding unit, the second de-quantization unit 435 in the decoding device 400 duplicates part or all 1024 samplings of the frequency spectrum data in the low-frequency band, with its 1024 samplings as the sampled data in the high frequency band.The frequency spectrum data of low-frequency band can extract similar data by the peak information according to frequency spectrum data in one or more scale factor from the frequency spectrum data of the first de-quantization unit, 430 outputs, then its some or all of method of duplicating is replicated.And if necessary, the amplitude of the frequency spectrum data that duplicates is adjusted in the second de-quantization unit 435.Amplitude can be by multiply by each frequency spectrum data a predetermined coefficient, and for example " 0.5 " is adjusted.This coefficient can be a fixing value, perhaps changes with each bandwidth or scale factor, and perhaps the frequency spectrum data of being exported with the first de-quantization unit 430 changes.

In the present embodiment, used a predetermined coefficient, but this coefficient value can be added in second coded message as sub-information.Perhaps scale factor value can be added in second coded message as a coefficient, and perhaps the quantized value of the peak value in the scale factor can be added in second coded message as a coefficient.Amplitude adjustment method is not limited to above-mentioned method, also can use other method.

In the present embodiment, have only positional information or have only positional information and coefficient information is encoded, but the present invention is not limited thereto.A scale factor, a quantized value, the symbolic information of frequency spectrum, noise production method etc. can be encoded.Perhaps the two or more combination in them also can be encoded.

In addition, in the present embodiment, the frequency spectrum data in the low-frequency band is replicated as the frequency spectrum data in the high frequency band.But the present invention is not limited thereto, and the frequency spectrum data in the high frequency band also can only produce from second coded message.

The spectrum waveform of the example of the sub-information (symbolic information) of other that accompanying drawing 19 has shown that second quantifying unit 345 shown in Figure 2 is produced.Accompanying drawing 20 is process flow diagrams, the operating process of sub-information (symbolic information) computing of other that shows that second quantifying unit 345 shown in Figure 2 is performed.

Second quantifying unit 345 is according to following processing procedure, specifies in to have to be higher than the precalculated position of 11.025kHz to each scale factor of the high frequency band of the regeneration bandwidth of 22.05kHz, for example in the symbolic information (S41) of the frequency spectrum data of central authorities.

The symbolic information (S42) of the frequency spectrum data of the middle position in first scale factor that second quantifying unit 345 is checked in the high frequency band with the reproduction band that is higher than 11.025kHz, and keep this value.For example, the symbol of the frequency spectrum data of the middle position in first scale factor is "+".Second quantifying unit 345 is with this symbol "+" of value representation of a bit " 1 " and keep it.When symbol was "-", second quantifying unit, 345 usefulness " 0 " expression also kept.

When the symbolic information of the frequency spectrum data of the middle position in first scale factor is maintained (S43), second quantifying unit 345 is checked the symbol (S42) of the frequency spectrum data of the middle position in the next scale factor.For example, symbol is "+", and second quantifying unit 345 keeps " 1 " and with its symbolic information as the frequency spectrum data of the middle position in second scale factor.

In an identical manner, the symbol "+" of the frequency spectrum data of the middle position of the 3rd scale factor in second quantifying unit, the 345 inspection high frequency bands, and hold mark information " 1 ".Second quantifying unit 345 is further checked the symbol "+" of the frequency spectrum data of the middle position in the 4th scale factor, and hold mark information is " 1 ".

When the symbolic information of the frequency spectrum data of the middle position of all proportions factor band in the high frequency band is all maintained (S43), second quantifying unit 345 is exported to the sub-information of second coding unit 355 as high frequency band with the symbolic information of the scale factor of maintenance, and this processing procedure finishes then.

As mentioned above, second quantifying unit 345 has produced sub-information (symbolic information).This sub-information (symbolic information) represents that with the symbolic information of a bit like this, the frequency spectrum of high frequency band just can utilize very short data length to represent with 4 scale factor in 1024 high frequency bands of representing of taking a sample of frequency spectrum data respectively.

In this case, the second de-quantization unit 435 of decoding device 400 duplicates the some or all of data of frequency spectrum data of 1024 samplings in the low-frequency band, frequency spectrum as high frequency band, according to the symbolic information of second decoding unit, 425 inputs, determine symbol then at the frequency spectrum data of a preposition.

The symbolic information of the symbol of the middle position of each scale factor in the expression high frequency band is used as sub-information (symbolic information).But the present invention is not limited to the middle position of scale factor, each peak, and first frequency spectrum data or other the precalculated position of each scale factor can be used.

In the present embodiment, scheduled with the position of the symbol that will be transmitted (symbolic information) corresponding frequency spectrum data, but it also can change according to the output of the first de-quantization unit 430, and the positional information of position that perhaps shows the symbolic information of each scale factor also be introduced into second coded message and is transmitted.

And if necessary, the amplitude of the frequency spectrum data that duplicates is adjusted in the second de-quantization unit 435.Amplitude is by each frequency spectrum data being multiply by a predetermined coefficient, and for example " 0.5 " is adjusted.

This coefficient can be a fixed value, perhaps can change with each bandwidth or scale factor, and perhaps the frequency spectrum data according to 430 outputs of the first de-quantization unit changes.The method of amplitude adjustment is not limited thereto, and other method also can be used.

In the present embodiment, used a predetermined coefficient, but this coefficient value be introduced in second coded message as sub-information.Perhaps scale factor value be introduced in second coded message as a coefficient, and perhaps the value of a quantification is added in second coded message as a coefficient.

In the present embodiment, symbolic information is only arranged, symbolic information and coefficient information are only arranged, perhaps only have symbolic information and positional information to be encoded, but the present invention is not limited thereto.The value of a quantification, a scale factor, the positional information of a characteristic frequency spectrum, a noise production method etc. can be encoded.Perhaps the two or more combinations in them can be encoded.

In addition, in the present embodiment, the frequency spectrum data of low-frequency band is replicated to the frequency spectrum data of high frequency band.But the present invention is not limited thereto, and the frequency spectrum data of high frequency band also can only produce from second coded message.

In the present embodiment, symbol "+" is with value " 1 " expression of 1 bit, and symbol "-" is represented with " 0 ".But the present invention is not limited to this expression of the symbol of (symbolic information) in the sub-information, also can use other value./

Accompanying drawing 21A and 21B have shown how expression produces the spectrum waveform figure of the example of other sub-information (Copy Info) that second quantifying unit 345 shown in Figure 2 produced.Accompanying drawing 21A has shown the spectrum waveform of first scale factor in the high frequency band.Accompanying drawing 21B has shown the example of the spectrum waveform of appointed low-frequency band with sub-information (Copy Info).Accompanying drawing 22 is process flow diagrams, shows the computing operating process of other sub-information (Copy Info) that second quantifying unit 345 shown in Figure 2 is performed.

Be higher than 11.025kHz each scale factor in the high frequency band of the regeneration bandwidth of 22.05kHz for having, second quantifying unit 345 is specified the number N (S51) of the scale factor in the low-frequency band according to following processing procedure.N scale factor in the low-frequency band is designated, because the value of the peak of this frequency band is near the peak " n " (from first " n " individual data that begin of scale factor) of the scale factor of high frequency band.

Second quantifying unit 345 is specified absolute maximum spectrum data (peak value) in first scale factor of the high frequency band with the regeneration bandwidth that is higher than 11.025kHz position " n " (S52).Shown in Figure 21 A, represent that 1. the peak value " n " of appointment and the frequency spectrum data value in this position are n=22.

Second quantifying unit 345 is specified the peak (S53) of all frequency spectrums in the low-frequency band with 11.025kHz or littler regeneration bandwidth (comprise on the occasion of with the frequency spectrum of negative value).

Next, each peak value for appointment in the low-frequency band, the search of second quantifying unit 345 is from the scale factor of its peak of first scale factor the most close " n ", and specifies number N, the direction of search and the symbolic information of peak value (S54) of this scale factor.

Particularly, for the peak value (comprising positive negative value) of each appointment in the low-frequency band, second quantifying unit 345 is from first scale factor of low-frequency band one side sequential search peak the most close " n ".The direction that two search are arranged: (1) from the peak value searching of low frequency direction and (2) from the peak value searching of high-frequency direction.In addition, for the peak value of low-frequency band, its sign symbol is come from the high frequency band counter-rotating, also exists two kinds of directions of search, and (3) are from the peak value searching of low frequency direction, and (4) are from the peak value searching of high-frequency direction.

In the direction of search is under the situation of (2) and (4), and when the spectrum waveform of low-frequency band was replicated according to peak information, the peak of high frequency band and the peak of low-frequency band passed through and be inverted (along the direction of frequency axis), shown in Figure 21 B.Therefore, need the information (forward direction and reverse) of the additional representation direction of search, for example, (1) and (3) is the sweep forward direction, and (2) and (4) are the reverse search direction.And, be under the situation of (3) and (4) in the direction of search, the peak of high frequency band and the peak of low-frequency band are by counter-rotating (along vertical axis) up and down, shown in Figure 21 B.Therefore, the information that needs the sign symbol of the peak value of additional representation high frequency band and low-frequency band whether to be inverted.

Second quantifying unit 345 is at the enterprising line search of four direction, be exactly, if the peak value of appointment is positive in low-frequency band, the direction of search is (1) and (2), if peak value is negative, the direction of search is (3) and (4), specifies peak in the Search Results near the number of the scale factor of " n " then.In this case, a specific value, for example " 5 " are pre the tolerance limit between " n " and actual peak location, and second quantifying unit 345 selects peak near the scale factor of " n " in four kinds of Search Results, and specify the number N of this scale factor.In addition, the information of also having specified the information whether symbol that shows the peak value in high frequency band and the low-frequency band be inverted and having shown the direction of search (forward direction or oppositely).

For example, along the direction of search (1), specifying the tolerance limit with peak for the frequency spectrum of low-frequency band is the scale factor of number N=3 of " 1 ", shown in Figure 21 B (1).Similar, in the direction of search (2), (3), (4) are specified the tolerance limit with peak for the frequency spectrum in the low-frequency band and are respectively " 5 ", number N=18 of " 4 " and " 2 ", N=12, the scale factor of N=10.Second quantifying unit 345 is selected tolerance limit with peak in four number purpose scale factor of these appointments be that " 1 " and peak are near number N=3 of the scale factor of " n ".In addition, produced the symbolic information " 1 " of the symbol "+" that shows the peak value in the low-frequency band and show the direction of search information " 1 " of in low-frequency band, searching for.In this case, if the symbol of peak value is "-", symbolic information is " 0 ", and if the search carry out in high frequency direction, direction of search information is " 0 ".

In the time of scale factor number N=3, for designated symbols information " 1 " in first scale factor in the high frequency band and direction of search information " 1 " (S55), second quantifying unit 345 is specified the number N of next scale factor in the same manner as described above, symbolic information and direction of search information.

By this way, the number N of each scale factor in the low-frequency band, symbolic information and direction of search information are designated, its apart from the most close high frequency band of peak of first frequency band from first peak that begins " n " of this scale factor (S55).Then, second quantifying unit 345 will with the number N of the appointment of scale factor in the corresponding low-frequency band of each scale factor of high frequency band, symbolic information and direction of search information are exported to the sub-information (Copy Info) of second coding unit 355 as high frequency band, and processing finishes then.

In this case, if first coded signal is decoded in decoding device 400 according to traditional program, the frequency spectrum data of 1024 samplings of low-frequency band side just can obtain.The second de-quantization unit 435 duplicate with the corresponding part of scale factor number of second decoding unit 425 output or all frequency spectrum datas as the frequency spectrum data in the high frequency band.If necessary, the amplitude of the frequency spectrum data that duplicates is adjusted in the second de-quantization unit 435.Amplitude is by each frequency spectrum being multiply by a predetermined coefficient, and for example " 0.5 " is adjusted.

Coefficient can be a fixed value, perhaps can be changed for each scale factor, and perhaps the frequency spectrum data along with 430 outputs of the first de-quantization unit can be changed.

In the present embodiment, used a predetermined coefficient, but this coefficient value be introduced into second coded message as sub-information.Perhaps scale factor value be introduced in second coded message as a coefficient, and perhaps quantized value be introduced in second coded message as a coefficient.And amplitude adjustment method is not limited to above-mentioned method, and any other method also can be used.

In the present embodiment, symbolic information and direction of search information, and the number N of scale factor is extracted the sub-information (Copy Info) as high frequency band.But according to transmissible information content in the high frequency band, symbolic information and direction of search information can be left in the basket.And when the peak value symbol of low-frequency band was "+", symbolic information was represented with " 1 ", and when symbol is "-", is represented with " 0 ".When search is a peak value from low frequency direction when carrying out, direction of search information is with " 1 " expression, and when search is a peak value from high frequency band direction when carrying out, it is with " 0 " expression.But the symbol of the peak value in the low-frequency band in symbolic information and the direction of search in direction of search information are not limited thereto, and they can be with other value representation.

And in the present embodiment, the peak from first appointment that begins in the low-frequency band is the most searched near first scale factor of " n ".But the present invention is not limited thereto, and the peak that first frequency band of the scale factor in the low-frequency band begins also can be searched near the peak value of " n ".

Accompanying drawing 23 shows a spectrum waveform figure who shows second example how setting up other sub-information (Copy Info) that second quantifying unit 345 shown in Fig. 2 produced.Accompanying drawing 24 is process flow diagrams, has shown the operating process of second kind of computing of other sub-information (Copy Info) that second quantifying unit 345 shown in Figure 2 is performed.

Be higher than 11.025kHz each scale factor in the high frequency band of the regeneration bandwidth of 22.05kHz for having, second quantifying unit 345 is according to following processing procedure, specify the number N of the scale factor in the low-frequency band, its be with high frequency band in poor (energy difference) of each frequency spectrum of scale factor be the frequency band (S61) of minimum.In this case, the number of the frequency spectrum data in the low-frequency band equals the number of the frequency spectrum data in the high frequency band, and the number N of the scale factor of appointment is represented first number of this scale factor.

For all proportions factor band (S62) in the low-frequency band, second quantifying unit 345 comprise with high frequency band in the frequency bandwidth of frequency spectrum data of scale factor similar number in, first data in the scale factor of low-frequency band are calculated poor (S63) of the frequency spectrum of the frequency spectrum of high frequency band and low-frequency band.For example, in waveform as shown in figure 23, if first scale factor of high frequency band comprises 48 samplings of frequency spectrum data, first data of the scale factor of number N=1 of second quantifying unit 345 in the low-frequency band, 48 frequency spectrum datas between order computation high frequency band and the low-frequency band poor.

When having calculated the difference of frequency spectrum data of high frequency band and low-frequency band when second quantifying unit 345, (S65), it keeps this value, then comprise with high frequency band in the frequency bandwidth of frequency spectrum data of scale factor similar number in, from first data of the next scale factor of low-frequency band, be poor (S64) that next scale factor is calculated the frequency spectrum between high frequency band and the low-frequency band.For example, in width in 48 samplings of frequency spectrum data, when beginning to calculate the difference of frequency spectrum from first data of the scale factor of number N=1 of low-frequency band, second quantifying unit 345 keeps the value of these differences that calculate, then further in the width of 48 samplings of frequency spectrum data first data of the scale factor of number N=2 in the low-frequency band begin to calculate the poor of frequency spectrum.In the same way, to from number N=3,4 ... all proportions factor band in the low-frequency band of 28 (last scale factor in the low-frequency band), second quantifying unit 345 is by sequentially calculating the poor of frequency spectrum with the difference summation of 48 frequency spectrum datas between high frequency band and the low-frequency band.

For all proportions factor band in the low-frequency band, second quantifying unit 345 with the width of the frequency spectrum data of high frequency band similar number in, first data of scale factor in the low-frequency band are calculated poor (S64) of the frequency spectrum between high frequency band and the low-frequency band.Then, second quantifying unit 345 is specified the difference that the calculates number N (S65) for minimum scale factor.For example, in spectrum waveform shown in Figure 23, the scale factor of number N=8 in the low-frequency band is just designated.In the figure, the difference that has shown the frequency spectrum data in the high frequency band of frequency spectrum data and dash area in the low-frequency band of dash area is for minimum, and the energy difference between the frequency spectrum also is a minimum.In other words, if 48 samplings of the frequency spectrum data that begins from first data of the scale factor of number N=8 are copied to first scale factor in the high frequency band that is higher than 11.025kHz, they have just become at the waveform of representing with the length dotted line that replaces in high frequency band shown in the accompanying drawing 23, therefore, the energy in the corresponding scale factor in high frequency band can be represented approx to original frequency spectrum.

Having specified the number in the low-frequency band when second quantifying unit 345 is the scale factor of N, when the difference of the frequency spectrum of the scale factor in itself and the high frequency band be minimum, it keeps the number of appointment is the scale factor of N, specified then with high frequency band in the corresponding low-frequency band of next scale factor in the number N (S66) of scale factor.Second quantifying unit 345 repeats this processing procedure in regular turn, and when the scale factor of having specified all number N in the low-frequency band, when the difference of the frequency spectrum of itself and high frequency band is minimum, the number N of the scale factor in the low-frequency band that is kept is outputed to second coding unit 355 as the sub-information (Copy Info) that is used for high frequency band, and this process finishes then.

In the present embodiment, the method for duplicating the method for the frequency spectrum data in the low-frequency band and adjusting its amplitude in decoding device 400 is with identical with reference to the situation of accompanying drawing 21 and 22 described sub-information (Copy Info).

In the process flow diagram of accompanying drawing 24, the energy difference of the same-sign of the frequency spectrum data between high frequency band and the low-frequency band is calculated on the equidirectional of frequency axis.But, encoding device of the present invention is not limited thereto, they can use one of three following methods to be calculated, as described in reference accompanying drawing 21 and 22: 1. for have same-sign in the high frequency band and on the direction of taking high frequency band from low frequency to by the frequency spectrum data of select progressively, the frequency spectrum data of similar number is sequentially selected in the low-frequency band, selection is first data of the scale factor in the low-frequency band, carry out along the direction of taking low-frequency band from high frequency to (reverse direction on the frequency axis), and the difference of frequency spectrum is calculated, 2. the symbol of the frequency spectrum in the low-frequency band is inverted (multiply by negative value) and is calculated on the equidirectional on the frequency axis, 3. the symbol of the frequency spectrum in the low-frequency band is inverted (multiply by negative value), and is calculated on the reverse direction of frequency axis.Perhaps, after having calculated the difference of energy, comprise in the low-frequency band that energy difference is that the scale factor of the number N of minimum frequency spectrum can be sub-information according to four kinds of all methods.Under the sort of situation, for low-frequency band self-energy difference is copied to high frequency band accurately for minimum frequency spectrum, for each scale factor, information that concerns between the symbol of the frequency spectrum of expression height frequency band and the information of duplicating direction that is illustrated on the frequency axis are inserted in the middle of the sub-information.The information that concerns between the symbol of the frequency spectrum of expression height frequency band represents with a bit, and for example, " 1 " expression has frequency spectrum poor of same-sign, and " 0 " expression has frequency spectrum poor of contrary sign.And, expression copies to the frequency spectrum data in the low-frequency band information of the direction on frequency axis of high frequency band and represents with a bit, for example, " 1 " expression forward direction duplicates direction, be exactly, select the forward direction of frequency spectrum data in the height frequency band, direction is oppositely duplicated in " 0 " expression, is exactly the inverse direction of selecting frequency spectrum data in the height frequency band.

As mentioned above, the situation that is applied to broadcast system according to the voice data distribution system of present embodiment is described.But it also can be applied to by the transmission medium such as the internet, with the form of bit stream, voice data is assigned to the voice data distribution system of terminal from client server.Perhaps, it also can be applied in such voice data distribution system, this voice data distribution system is in case the bit stream that encoding device 300 is exported is recorded in a kind of such as the CD that comprises CD and DVD, semiconductor, perhaps on the recording medium of a hard disk, just by this recording medium in decoding device 400 with its regeneration.

In the present embodiment, processing utilizes a LONG piece to be performed, but it also can use a SHORT piece to carry out.Use a SHORT piece can carry out the processing same with the LONG piece.

In encoding process, such as gain control, TNS (instantaneous noise moulding), acoustic module at heart, M/S is stereo, intensity stereo and prediction, the variation of module size, instruments such as bit deposit can be used.

In the present embodiment, sub-information is to produce according to the frequency spectrum data of being divided by data division unit 330 in the high frequency band.But sub-information also can produce according to the resulting value of de-quantization is carried out in the output of first quantifying unit 340, as the frequency spectrum data in the high frequency band.

In the present embodiment, be used to obtain the scale factor of the quantized value " 1 " of the frequency spectrum data in each scale factor in high frequency band, the value of quantification, the positional information of characteristic frequency spectrum shows that the symbolic information etc. of the sign symbol of frequency spectrum all is used as sub-information.But the combination of two or more in them also can be used as sub-information.In this case, if scale factor and expression gain, being combined in of coefficient of the position of the frequency spectrum data of absolute value maximum etc. is encoded in the sub-information, and it is very effective so.And, in the present embodiment, for each scale factor, a sub-information can be encoded as second coded signal, but a sub-information also can be encoded and be used for two or more scale factor, and perhaps two or more sub-information are encoded and are used for a scale factor.In addition, in this enforcement, can be each chnnel coding sub-information of encoding, perhaps a sub-information is encoded and is used for two or more channels.

In the present embodiment, encoding device 300 comprises two quantifying unit and two coding units.But the present invention is not limited thereto, and it also can comprise three or a plurality of quantifying unit and coding unit respectively.

In the present embodiment, decoding device 400 comprises two decoding units and two de-quantization unit.But the present invention is not limited thereto, and it also can comprise three or a plurality of decoding unit reconciliation quantifying unit respectively.

Above-described processing procedure can also realize that except realizing with hardware the present invention can be configured to a part of processing and utilizing hardware and realize that other processing and utilizing software is realized with software.

In the present embodiment, the sample frequency of use is 44.1kHz, but other sample frequency, and for example 32kHz or 48kHz can be used.As data division unit 330, the frequency of the boundary of division frequency spectrum data can become other any frequency except 11.025kHz.

And in the present embodiment, processing is performed according to MPEG-2 AAC.But same processing also can be according to other method (for example MP3, AC3 etc.) at an encoding device, a decoding device, and be performed in the miscellaneous equipment.

And, also can adopt following structure according to encoding device of the present invention.

Encoding device according to the present invention is the encoding device of a coding audio data, comprise: a separative element is used for a voice data string is separated into m2 sampling, more than m1 sampling of requisite number purpose, separate continuous voice data from the voice data string that produces; A converter unit is used for the voice data that separative element separates is transformed into the frequency spectrum data of frequency field; A division unit, the m2 sampling that is used for frequency spectrum data that conversion is obtained be divided into that m1 of the frequency spectrum data in the low-frequency band takes a sample and high frequency band in (m2-m1) individual sampling of frequency spectrum data; A low-frequency band coding unit is used to the data of the frequency spectrum data that quantizes to divide and this quantification of encoding in low-frequency band; A sub-information generation unit, the frequency spectrum data that is used to quantize the division in the high frequency band produces the sub-information that shows the spectral characteristic in the high frequency band; A high frequency band coding unit is used to the sub-information of encoding and producing; And an output unit, what be used for that sign indicating number that the low-frequency band coding unit is obtained and high frequency band coding unit obtained is yard integrated, and exports this integrated symbol.

In this case, thereby sub-information generation unit can dispose like this and can calculate a normalization factor that is used to obtain a fixed value, and the normalization factor that generation is calculated is as sub-information, aforesaid fixed value is a value that obtains by the peaks spectrum data that quantize to be used for each group in the frequency data high frequency band, and wherein frequency spectrum data has been divided into a plurality of groups.

And, thereby sub-information generation unit can dispose like this can use to all common normalization factor of each group quantize the peaks spectrum data in each group in the high frequency band and value of produce quantizing as sub-information, wherein frequency spectrum data has been divided into a plurality of groups.

And, thereby sub-information generation unit can dispose like this in each group in high frequency band and produces peaks spectrum data frequency position as sub-information, and wherein frequency spectrum data has been divided into a plurality of groups.

And, frequency spectrum data is a MDCT coefficient, thereby sub-information generation unit can dispose the symbol of the positive negative value that produces a frequency spectrum data that is illustrated in a predetermined frequency location in the high frequency band like this as sub-information, and wherein frequency spectrum data has been divided into a plurality of groups.

Further, thus sub-information generation unit can dispose like this produce an appointment near the information of the frequency spectrum in the low-frequency band of the frequency spectrum of each group in the high frequency band as sub-information, wherein frequency spectrum data has been divided into a plurality of groups.In this case, thereby sub-information generation unit can dispose the frequency spectrum of specifying in the low-frequency band like this, wherein on frequency axis on the distance of the spectrum peak of the delimiter of the group in high frequency band to this group and the frequency axis distance of the peak value of the frequency spectrum of the delimiter of the group in low-frequency band to this group be minimum.And, thereby sub-information generation unit can dispose a frequency spectrum of specifying in the low-frequency band like this, its with high frequency band in frequency spectrum in this group to have the energy difference that identical band width obtains be minimum.And the information of specifying the frequency spectrum in the low-frequency band is the numeral of the group of a frequency spectrum of specifying the appointment in the low-frequency band.

And, thereby sub-information generation unit can be configured to produce a predetermined coefficient of the spectrum amplitude gain in the expression high frequency band as sub-information.

And, output unit may further include a stream output unit, be used for low-frequency band coding unit coded data is transformed into the coded audio stream with predetermined formal definition, then high frequency band coding unit coded data is stored in its zone of using the coded audio stream that is not subjected to the coding protocol restriction, exports the data of this storage at last.In this case, thus the stream output unit can be configured to write the information of expression f1Hz as sample frequency.

Further, output unit may further include one second stream output unit, be used for low-frequency band coding unit coded data is transformed into the coded audio stream with predetermined formal definition, then high frequency band coding unit coded data is stored in coded audio and flows not in the homogeneous turbulence, export the data of this storage then.

It should be noted that, the present invention can realize with a kind of encoding device of above-mentioned modification and communication system of decoding device of comprising certainly, also can realize with a kind of coding method and the communication means of step performed in the feature unit that is included in above-mentioned encoding device and communication system, realize with a kind of feature unit that CPU carries out above-mentioned encoding device or coded program of step of making, perhaps realize with a kind of computer readable recording medium storing program for performing that is writing down these programs on it.

Commercial Application

Encoding device according to the present invention is suitable for use as a kind of distribution system and is used for being distributed in data flow In or in the recording medium such as contents such as music.

Claims (27)

1. encoding device that is used for coding audio data comprises:

A separative element is used for the voice data string is separated into the continuant audio data of a fixed number;

A converter unit, the voice data that is used for separating is transformed into the frequency spectrum data of frequency field;

A division unit is used for that the frequency spectrum data that converter unit obtains is divided into the frequency spectrum data in frequency f 1Hz or lower low-frequency band and is being higher than frequency spectrum data in the high frequency band of f1Hz;

A low-frequency band coding unit is used to quantize frequency spectrum data and this quantized data of encoding of the division in the low-frequency band;

A sub-information generation unit is used to produce the sub-information of the spectral characteristic of the high frequency band that shows the frequency spectrum data that comes from the division in the high frequency band;

A high frequency band coding unit is used to the sub-information of encoding and producing;

And an output unit, be used for integrated sign indicating number that obtains by the low-frequency band coding unit and the sign indicating number that obtains by the high frequency band coding unit, and export integrated sign indicating number,

Wherein f1 be sample frequency f2 half or littler, produce the voice data string at frequency f 2 places.

2. encoding device as claimed in claim 1,

Wherein f1 is f2/4,

Converter unit is transformed into the frequency spectrum data of 0～2 * f1Hz with voice data, and

Division unit with the frequency spectrum data of 0～2 * f1Hz be divided in frequency f 1Hz or lower low-frequency band frequency spectrum data and at the frequency spectrum data in the high frequency band of 2 * f1Hz more than the f1Hz.

3. encoding device as claimed in claim 2,

Wherein the frequency spectrum data in frequency f 1Hz or lower low-frequency band comprises n frequency spectrum data sampling,

Separative element can be separated into the voice data string and produce 2 * n frequency spectrum data sampling requisite number purpose voice data,

Converter unit is transformed into 2 * n frequency spectrum data sampling with the voice data that separates,

Division unit is taken a sample n the frequency spectrum data that n frequency spectrum data taken a sample and high frequency band is interior that 2 * n frequency spectrum data sampling is divided in the low-frequency band.

4. encoding device as claimed in claim 3,

Wherein separative element is separated into 2 * n frequency spectrum data sampling with the voice data string, the sampling of these frequency spectrum datas comprises with the corresponding n of a frame voice data sampling and two groups of n/2 voice datas before this frame and in two frames of vicinity afterwards as a coding unit takes a sample

Converter unit is carried out MDCT to 2 * n the voice data sampling that separates, and it is transformed into the frequency spectrum of 0～2 * f1Hz that comprises 2 * n frequency spectrum data sampling.

5. encoding device as claimed in claim 1,

Its neutron information generation unit calculates a normalization factor that is used for deriving a fixed value that obtains by the peaks spectrum data that quantize each group in high frequency band, and produced a normalization factor of calculating gained, as sub-information, wherein frequency spectrum data has been divided into a plurality of groups.

6. encoding device as claimed in claim 1,

Its neutron information generation unit utilization is common normalization factor for described each group, quantizes the peaks spectrum data in each group in high frequency band, and produces quantized value as sub-information, and wherein frequency spectrum data has been divided into a plurality of groups.

7. encoding device as claimed in claim 1,

Its neutron information generation unit produced in high frequency band each the group in peaks spectrum data frequency position as sub-information, frequency spectrum data has been divided into a plurality of groups.

8. encoding device as claimed in claim 1,

Wherein frequency spectrum data is a MDCT coefficient,

The symbol of the positive negative value of the predetermined frequency location generation expression frequency spectrum data of sub-information generation unit in high frequency band is as sub-information, and frequency spectrum data has been divided into a plurality of groups.

9. encoding device as claimed in claim 1,

Its neutron information generation unit produced an appointment near the information of the frequency spectrum in the low-frequency band of the frequency spectrum in each group in the high frequency band as sub-information, frequency spectrum data has been divided into a plurality of groups.

10. encoding device as claimed in claim 9,

Its neutron information generation unit is specified the frequency spectrum in low frequency bands of a spectrum, and wherein to accord with the difference between the distance of the spectrum peak in this group be minimum to the boundary that accords with each group in the low-frequency band in the distance of the spectrum peak in this group and (2) frequency field, the boundary of (1) each group in frequency field medium-high frequency band.

11. encoding device as claimed in claim 9,

Its neutron information generation unit is specified the frequency spectrum in low-frequency band, and wherein the difference of the energy value that obtains in the frequency bandwidth identical with frequency spectrum in this group in the high frequency band is a minimum.

12. encoding device as claimed in claim 9,

The information of wherein specifying the frequency spectrum in the low-frequency band is a numeral that specifies in the group of the assigned frequency in the low-frequency band.

13. encoding device as claimed in claim 1,

Its neutron information generation unit has produced a predetermined coefficient as sub-information, shows the amplitude gain of the frequency spectrum in the high frequency band.

14. encoding device as claimed in claim 1,

Wherein output unit further comprises a stream output unit, be used for the coded audio that low-frequency band coding unit coded data is transformed to predetermined formal definition is flowed, high frequency band coding unit coded data is stored in it uses the zone in the unrestricted coded audio stream under predetermined format, then the data of storage are exported.

15. encoding device as claimed in claim 14,

Wherein flow output unit and write the information of expression f/2Hz as sample frequency.

16. encoding device as claimed in claim 1,

Wherein output unit also further comprises one second stream output unit, be used for the coded audio that low-frequency band coding unit coded data is transformed to predetermined formal definition is flowed, high frequency band coding unit coded data is stored in the data stream different with coded audio stream, then the data of storage is exported.

17. the decoding coded data by a recording medium or transmission medium input

Decoding device comprises:

An extracting unit is used for extracting the coded data of the low-frequency band that is included in coded data and the coded data of high frequency band;

A low-frequency band de-quantization unit is used to decode and the coded data of the low-frequency band that de-quantization is extracted by extracting unit, thereby output is at the frequency spectrum data of frequency f 1Hz or lower low-frequency band;

A sub-information decoding unit, the coded data of the high frequency band that the extracting unit that is used to decode extracts, thus produce the sub-information that shows the frequency spectrum data characteristic in the high frequency band;

A high frequency band de-quantization unit is used for the sub-information according to the generation of sub-information decoding unit, the frequency spectrum data in the output high frequency band;

An integrated unit is used for the frequency spectrum data of integrated low-frequency band by low-frequency band de-quantization unit output and by the frequency spectrum data in the high frequency band of high frequency band de-quantization unit output;

An inverse transformation block is used for and will be changed into the voice data of time domain by the integrated frequency spectrum data inversion of integrated unit;

A voice data output unit is used for based on the voice data of time sequencing output by the inverse transformation block inverse transformation.

18. decoding device as claimed in claim 17,

Its neutron information is the normalization factor of a calculating, so that the value that the peaks spectrum data in quantification each group in high frequency band are obtained all is a fixed value, frequency spectrum data has been divided into a plurality of groups, and,

Quantized value of normalization factor de-quantization in the sub-information that high frequency band de-quantization unit by using produces, this quantized value be for described each group in the high frequency band be common and with described each group in the corresponding fixed value of frequency spectrum data of preset frequency, produce frequency spectrum data then in high frequency band, its peak value is the frequency spectrum data of the de-quantization in described each group.

19. decoding device as claimed in claim 17,

Its neutron information is that utilization is common normalization factor to described each group, quantizes the quantized value that the peaks spectrum data in each group in high frequency band are obtained, and frequency spectrum data has been divided into a plurality of groups, and,

High frequency band de-quantization unit by using is common normalization factor to described each group, the quantized value of de-quantization in the sub-information that produces, and produce the interior frequency spectrum data of high frequency band, its peak value is the frequency spectrum data of the de-quantization in described each group.

20. encoding device as claimed in claim 17,

Its neutron information is the peaks spectrum data frequency position in each interior group of high frequency band, and frequency spectrum data has been divided into a plurality of groups, and

High frequency band de-quantization unit produces frequency spectrum data in high frequency band, its peak value is arranged in the frequency location of described each sub-information of being produced of group.

21. decoding device as claimed in claim 17,

Its neutron information is meant the symbol of the positive negative value of the frequency spectrum data that is shown in the predetermined frequency location in the high frequency band, and frequency spectrum data has been divided into a plurality of groups, and

High frequency band de-quantization unit is created in the frequency spectrum data of the preset frequency position in the high frequency band, and it has the symbol in the sub-information that produces.

22. decoding device as claimed in claim 17,

Its neutron information is the information of specifying the most near the frequency spectrum in the low-frequency band of the frequency spectrum of each group of high frequency band, and frequency spectrum data has been divided into a plurality of groups, and,

High frequency band de-quantization unit produces a predetermined noise according to sub-information in described each group of high frequency band, also join described frequency spectrum data by the noise that will produce and produce frequency spectrum data in the high frequency band.

23. one kind is equipped with the voice data distribution system that low bit rate was compressed and was encoded into the voice data of a bit stream by recording medium or transmission medium branch, this system comprises an encoding device and a decoding device:

Encoding device coding audio data wherein, and comprise:

A separative element is used for the voice data string is separated into the continuant audio data of a fixed number;

A converter unit, the voice data that is used for separating is transformed into the frequency spectrum data of frequency field;

A division unit is used for that the frequency spectrum data that converter unit obtains is divided into the frequency spectrum data in frequency f 1Hz or lower low-frequency band and is being higher than frequency spectrum data in the high frequency bandwidth of f1Hz;

A low-frequency band coding unit is used to quantize frequency spectrum data and this quantized data of encoding of the division in the low-frequency band;

A sub-information generation unit is used to produce the sub-information of the spectral characteristic of the high frequency band that shows the frequency spectrum data that comes from the division in the high frequency band;

A high frequency band coding unit is used to the sub-information of encoding and producing;

An output unit is used for integrated sign indicating number that is obtained by the low-frequency band coding unit and the sign indicating number that is obtained by the high frequency band coding unit, and exports integrated sign indicating number,

Wherein f1 be sample frequency f2 half or littler, produce the voice data string at frequency f 2 places, and

The coded data that the decoding device decoding is imported by recording medium or transmission medium, and comprise:

An extracting unit is used for extracting the coded data of the low-frequency band that is included in coded data and the coded data of high frequency band;

A low-frequency band de-quantization unit is used to decode and low-frequency band coded data that de-quantization is extracted by extracting unit, thereby output is at the frequency spectrum data of frequency f 1Hz or lower low-frequency band;

A sub-information decoding unit, the high frequency band coded data that is used to decode and extracts by extracting unit, thus produce the sub-information of the frequency spectrum data characteristic that shows high frequency band;

A high frequency band de-quantization unit is used for according to the sub-information by the generation of sub-information decoding unit, the frequency spectrum data in the output high frequency band;

An integrated unit is used for the frequency spectrum data of integrated low-frequency band by low-frequency band de-quantization unit output and by the frequency spectrum data in the high frequency band of high frequency band de-quantization unit output;

An inverse transformation block is used for and will be changed into the voice data of time domain by the integrated frequency spectrum data inversion of integrated unit;

A voice data output unit is used for the voice data based on the inverse transformation of time sequencing output inverse transformation block.

24. a program that is used for the encoding device of coding audio data, this program make the following effect of computer realization:

A separative element is used for audio data stream is separated into the continuant audio data of a fixed number;

A converter unit, the voice data that is used for separating is transformed into the frequency spectrum data of frequency field;

A division unit is used for being divided into the frequency spectrum data that converter unit obtains at the frequency spectrum data of frequency f 1Hz or lower low-frequency band and being higher than frequency spectrum data in the high frequency bandwidth of f1Hz;

A low-frequency band coding unit is used to quantize frequency spectrum data and this quantized data of encoding of the division in the low-frequency band;

A sub-information generation unit is used to produce the sub-information of the spectral characteristic of the high frequency band that shows the frequency spectrum data that comes from the division in the high frequency band;

A high frequency band coding unit is used to the sub-information of encoding and producing;

And an output unit, be used for integrated sign indicating number and the sign indicating number that obtains by the high frequency band coding unit, this integrated sign indicating number of output then that obtains by the low-frequency band coding unit.

25. the program of the decoding device of the coded data by recording medium or transmission medium input that is used for decoding, this program makes the following effect of computer realization:

An extracting unit is used for extracting the coded data of the low-frequency band that is included in coded data and the coded data of high frequency band;

A low-frequency band de-quantization unit is used to decode and the coded data of the low-frequency band that de-quantization is extracted by extracting unit, thereby output is at the frequency spectrum data of frequency f 1Hz or lower low-frequency band;

A sub-information decoding unit, the coded data of the high frequency band that the extracting unit that is used to decode extracts, thus produce the sub-information of the frequency spectrum data characteristic that shows high frequency band;

A high frequency band de-quantization unit is used for the sub-information according to the generation of sub-information decoding unit, the frequency spectrum data in the output high frequency band;

An integrated unit is used for the frequency spectrum data of low-frequency band of integrated low-frequency band de-quantization unit output and the high frequency band frequency spectrum data of high frequency band de-quantization unit output;

An inverse transformation block is used for the voice data that the frequency spectrum data inversion that integrated unit is integrated changes time domain into;

A voice data output unit is used for based on the inversion of time sequencing output inverse transformation block

The voice data that changes.

26. a computer-readable recording medium that is writing down program on it,

Wherein program is used for the encoding device of coding audio data, and this program makes the following effect of computer realization:

A separative element is used for audio data stream is separated into the continuant audio data of a fixed number;

A converter unit, the voice data that is used for separating is transformed into the frequency spectrum data of frequency field;

A division unit is used for that the frequency spectrum data that converter unit obtains is divided into the frequency spectrum data in frequency f 1Hz or lower low-frequency band and is being higher than frequency spectrum data in the high frequency bandwidth of f1Hz;

A low-frequency band coding unit is used to quantize frequency spectrum data and this quantized data of encoding of the division in the low-frequency band;

A sub-information generation unit is used to produce the sub-information of the spectral characteristic of the high frequency band that shows the frequency spectrum data that comes from the division in the high frequency band;

A high frequency band coding unit is used to the sub-information of encoding and producing;

And an output unit, be used for sign indicating number and the sign indicating number of high frequency band coding unit acquisition, this integrated sign indicating number of output then that integrated low-frequency band coding unit obtains.

27. a computer-readable media that is writing down program on it,

The program decoding device of the coded data by recording medium or transmission medium input that is used to decode wherein, this program makes the following effect of computer realization:

An extracting unit is used for extracting the coded data of the low-frequency band that is included in coded data and the coded data of high frequency band;

A low-frequency band de-quantization unit is used to decode and low-frequency band coded data that de-quantization is extracted by extracting unit, thus the frequency spectrum data of output in frequency f 1Hz or lower low-frequency band;

A sub-information decoding unit, the high frequency band coded data that is used to decode and extracts by extracting unit, thus produce the sub-information of the frequency spectrum data characteristic that shows high frequency band;

A high frequency band de-quantization unit is used for the sub-information according to the generation of sub-information decoding unit, the frequency spectrum data in the output high frequency band;

An integrated unit is used for the frequency spectrum data of integrated low-frequency band by low-frequency band de-quantization unit output and by the high frequency band frequency spectrum data of high frequency band de-quantization unit output;

An inverse transformation block is used for the voice data that the frequency spectrum data inversion that integrated unit is integrated changes time domain into;

A voice data output unit is used for based on the voice data of time sequencing output by the inverse transformation block inverse transformation.

Images